Abstract Aims Reactive oxygen species (ROS) play an important role in the pathogenesis of pulmonary arterial hypertension (PAH) and NADPH oxidases (NOXs) as sources of ROS are implicated in the development of the disease. We previously showed that NOX isozyme 1 (NOX1)-derived ROS contributes to pulmonary vascular endothelial cell (EC) proliferation in response to PAH triggers in vitro. However, whether and how NOX1 is involved in PAH in vivo have not been explored nor has NOX1 been examined as a viable and effective therapeutic disease target. Methods and results Herein, infusion of mice exposed to Sugen/hypoxia (10 % O[2]) with a specific NOX1 inhibitor, NOXA1ds, delivered via osmotic minipumps (i.p.), significantly suppressed pathological changes in hemodynamic parameters characteristic of PAH. Furthermore, lungs of human patients with idiopathic PAH (iPAH) and exploratory RNA-seq analysis of hypoxic human pulmonary ECs, in which NOX1 was suppressed, were probed. The findings showed a clear indication of NOX1 in the promotion of both protein disulfide isomerase (PDI) and the unfolded protein response (UPR; in particular, the PERK arm of the pathway including eIF2α and ATF4) leading to proliferation. In aggregate, these results are consistent with a causal role for NOX1 in the development of mouse and human PAH and reveal a novel and mechanistic pathway by which NOX1 activates the UPR response during EC proliferation. Conclusion NOX1 promotes phenotypic changes in ECs that are pivotal to proliferation and PAH through activation of the UPR. Taken together, our results are consistent with selective inhibition of NOX1 as a novel modality for attenuating PAH. Keywords: NADPH oxidases, NOXA1ds, NOX inhibitors, Hypoxia, Unfolded protein response, Pulmonary arterial hypertension Graphical abstract [43]Image 1 [44]Open in a new tab 2. Introduction Pulmonary arterial hypertension (PAH) is a progressive debilitating disorder characterized by the sustained elevation in the mean pulmonary artery pressure (mPAP > 20 mmHg) of various etiologies characterized by the presence of pulmonary vascular remodeling and obstructive lesions that ultimately lead to right ventricular failure [[45]1]. Although singular and combinatorial vasodilatory treatments have allowed for major improvements in the 5-year survival rate of many PAH patients in recent years, the root cause of PAH and curative therapies that reverse the disease are still elusive [[46][2], [47][3], [48][4]]. As such, a poor understanding of pathognomonic mechanisms driving PAH is a barrier to treatment. To address this, new mechanistic insight as well as new drug target identification are needed. In rodent models, PAH [[49]5,[50]6] is linked to increased reactive oxygen species (ROS) production; notably, those derived from NADPH oxidases. However, the role of these ROS is not fully understood [[51]7,[52]8]. The NADPH oxidases (NOXs) are a family of professional ROS-producing enzymes comprised of seven homologous core hemoproteins (NOX1-5 and DUOX1/2) that, depending on the isoform, might require assembly of additional subunits for their enzymatic competency in ROS production [[53]9]. Increased ROS production is linked to aggressive endothelial [[54][10], [55][11], [56][12], [57][13]] (EC) and smooth muscle cell [[58][14], [59][15], [60][16]] (SMC) proliferation and migration which are hallmarks of vascular remodeling in PAH and play seminal roles in propagation of the disease. Specifically, our work has focused on the observation that hypoxia-induced NOX1-derived ROS drive pulmonary EC proliferation and migration [[61][11], [62][12], [63][13]]. The unfolded protein response (UPR) is widely considered a consequence of ER stress as a result of a markedly increased number of misfolded proteins but is also viewed as a mechanism by which cells can promote cellular survival [[64]17,[65]18]. UPR involves activation of one or more of the three ER-anchored transmembrane receptors: inositol-requiring enzyme-1α (IRE1α), protein kinase R-like endoplasmic reticulum kinase (PERK), and/or activating transcription factor 6 (ATF6) [[66]19]. While much discussion in the literature focuses on the role of ER stress and ROS in cardiovascular disease [[67][20], [68][21], [69][22]], little to no connection has been drawn between NOX/ROS and UPR in PAH, no less in the broader disease classifications of PH. Importantly, despite a commonly drawn similarity between the EC proliferative phenotype in PAH and tumorigenesis [[70][23], [71][24], [72][25]] and an appreciation for the role of UPR in cancer [[73][26], [74][27], [75][28], [76][29]], the interconnectedness of NOX1 with UPR in PAH, is highly warranted. And, while association of NOX subunits, i.e. p22^phox [[77]30], NOX2 [[78]31,[79]32], and NOX4 [[80]32,[81]33] with UPR has been implied in some cardiovascular cells and disease states, NOX, and more specifically NOX1, as an upstream driver of the UPR response in PAH, and more pointedly EC proliferation/dysfunction, has not been proposed. Therefore, we postulated that EC NOX1 promotes UPR activation and leads to EC hyperproliferation, thus contributing to the PAH phenotype. In this study, we employed the Sugen-Hypoxia in vivo mouse model of PAH, which combines mouse exposure to chronic hypoxia and VEGF receptor antagonist (Sugen) administration and together instigate EC damage and hyperproliferation of remaining ECs [[82]34,[83]35]. Human lung ECs and human PAH tissue were used in vitro to interrogate our hypothesis. Herein, we show that the infusion of mice with a specific NOX1 inhibitor, NOXA1ds, attenuated hemodynamic parameters characteristic of PAH. Furthermore, RNA-seq analysis of hypoxic human pulmonary ECs reveal NOX1's role in the promotion of UPR and EC proliferation. Moreover, empirical findings showed that NOX1-upregulated UPR signaling via the PERK pathway is causally involved in the EC proliferation and this pathway was substantiated in human patients with idiopathic PAH. Thus, our data appear to have deciphered a novel pathway by which NOX1 activates the UPR response resulting in EC proliferation and PAH. Taken together, our results indicate that selective inhibition of NOX1 is a potential therapeutic strategy for PAH. 3. Materials and methods 3.1. Chemicals All chemicals were from Sigma-Aldrich unless stated differently. 3.2. Cell culture Human pulmonary artery endothelial cells (HPAEC) were obtained from Lonza (CC-2530) and cultured in EBM-2 medium containing EGM-2 components (Lonza, CC-3162) at 5 % CO[2], 37°C. Lot numbers of HPAECs obtained from Lonza varied amongst experiments and included lots: 23TL253310 (61 year old male, 61 M), 22TL073039 (59 M), 23TL170127 (61 M), 19TL170887 (60F), and 21TL228096 (56F) (certificates of analysis can be found online). HPAECs were pooled and experimented upon from passages 3–8. 3.3. Hypoxia Cells (HPAEC) were treated (as per Transfections and gene silencing), serum starved (1 % FBS) for 16 h. Cells were placed in an hypoxic chamber (BioSpherix, OxyCycler, model C42) incubators exposed to 1 % O[2,] 5 % CO[2], 37°C for 24 h or under normobaric conditions exposed to 21 % % O[2], 5 % CO[2], 37°C for 24 h. 3.4. Transfections and gene silencing For gene silencing, HPAECs were transfected with short interfering RNAs (siRNA) as previously described [[84]11,[85]36]. In short, siRNA targeting either NOX1 (ThermoFisher, Assay ID: s25726 [UniGene ID: Hs.592227] and s25728 [UniGene ID:Hs.592227]) or ATF4 (ThermoFisher, Assay ID: 122168 [UniGene ID: [86]Hs496487]) were used at a final concentration of 10 nM according to the manufacturer's protocol (ThermoFisher, Lipofectamine RNAiMax, 13778150). As a transfection control (Ctr) scrambled RNA was used (ThermoFisher). For transfection, HPAECs were seeded at a 60–70 % density and transfected on the following day with NOX1, ATF4 siRNA (10 nmol/l) or scrambled control siRNA (10 nmol/l) using Lipofectamine RNAiMax (ThermoFisher, Lipofectamine RNAiMax, 13778150) according to the manufacturer's protocol. Cells were serum starved (1 % FBS) for 16 h and subsequently subjected to normoxic (Nx; 21 % O[2]) or hypoxic (Hx; 1 % O[2]) conditions. Gene silencing was confirmed by Western blot displaying a 60–80 % knockdown. 3.5. Animals All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pittsburgh and are in accordance with the National Institutes of Health guidelines. Wild type (WT) male C57BL/6J mice (8 wks old) were purchased from the Jackson Laboratories (Bar Harbor, Me). Osmotic minipumps (Alzet 1004, filling volume 100 μl, delivery rate 0.11 μl/h over 3 wks) containing either scrambled peptide (Scr, 20 mg/kg/day, Sequence: [H]-LMKGPDAEKVA-[NH2], Tufts University, Department of Immunology) or NOXA1ds (20 mg/kg/day, Sequence: [H]-EPMDALGKAKV-[NH2], Tufts University, Department of Immunology) were implanted intraperitoneally (IP) in mice under sterile conditions on day -1. The following day (day zero) mice were placed either under normoxic (Nx, 21 % O[2]) or hypoxic (Hx, 10 % O[2]) conditions (BioSpherix, A-Chamber animal chambers, ProOx 110 oxygen controllers) for a duration of 21 d, thus, creating four distinct experimental conditions (Normoxia/Scrambled, Hypoxia/Scrambled, Normoxia/NOXA1ds, Hypoxia/NOXA1ds). Simultaneously, on days 0, 7, and 14 normoxic mice were subcutaneously injected with 200 μl of vehicle solution (20 % DMSO, 30 % Tween 80, 50 % 0.9 % NaCl), while hypoxic mice received 200 μl of Sugen 5416 solution (20 mg/kg Sugen 5416 dissolved in the vehicle). After the 21 d treatment, mice were anesthetized by 5 % isoflurane and open chest catheterization for PV-loops was performed (see Hemodynamic Measurements). 3.6. Hemodynamic measurements Mice were briefly anesthetized with 5 % isoflurane to allow restraint, placed on a warming pad, and exposed to the continuous nasal delivery of isoflurane (2.5 %). A rectal thermo-probe was inserted to monitor and ensure a stable body temperature of 37°C. Following the onset of deep surgical anesthesia, a tracheotomy was performed and a breathing tube was inserted into the trachea and attached to a ventilator. The thoracotomy was performed to allow free access to the heart. Anesthesia was then reduced to 1 % isoflurane. A 1.2 French catheter (Scisense) was inserted into the right ventricle (RV) and the pressure profile recorded after stabilization for 1 min. The catheter was then carefully advanced to the left ventricle (LV). The correct position was confirmed by observing the characteristic ventricular waveforms. Pressure profiles were recorded after stabilization for 1 min. LV and RV pressures (systolic, diastolic), LV and RV peak rate of pressure rise during systole (contraction, LV or RV dP/dt max), LV and RV peak rate of pressure decline during diastole (relaxation, LV or RV dP/dt min), mean pulmonary pressure (mPAP), pulmonary vascular resistance (PVR) Woods, RV stroke work, LV contractile index (LVCI) and aortic mean pressure (AOMP) were determined from pressure volume (PV) loops. Following the PV loop measurements, animals were sacrificed via CO[2]/O[2] exposure followed by cervical dislocation and transcardial perfusion was performed by placing a 25G infusion needle in the LV and making a small incision in the left atrium. PBS was allowed to perfuse the mouse under pressure of approximately 15-25 mmHg (∼250-300 μL/s) until tissue ran clear of blood and tissues were then harvested and placed on ice until the time of homogenization or pressure-fixed with 4 % paraformaldehyde (PFA) for histology/immunofluorescent imaging. 3.7. Protein isolation For protein isolation from HPAECs, murine lung tissue, and autopsied iPAH patient lungs (see human specimen collection), 1x lysis buffer (0.2 Tris, 1.5 M NaCl, 0.1 mM EGTA, 0.025 mM Na[4]P[2]O[7], 0.01 mM β-glycerophosphate, 0.01 M Na[3]VO[4], 1.5 % Triton X) was used. HPAECs were scraped into 1x ice-cold lysis buffer. Mouse and autopsied iPAH patient lungs were harvested, macerated/homogenized in 200 μl ice-cold lysis buffer in a glass-on-glass mortar and pestle immersed in ice. Homogenized lung tissue was passed through a 30-gauge needle six times for further disruption and homogenate consistency. Homogenates (aka lysates) from both cell and tissue were further dispersed in lysis buffer on a rocking table in a cold room overnight at 4°C. Lysates were then centrifuged at 15,000×g (at 4°C) and supernatants were collected. Subsequently, supernatant protein concentration was determined by Bradford protein assay. 3.8. Measurements of NADPH-mediated H[2]O[2] production by Amplex Red assay Ten micrograms of protein (per well) isolated from either murine lung tissue, or autopsied iPAH patient lungs were dissolved in Amplex Red assay mixture (25 mM HEPES pH 7.6, containing: 120 mmol/l NaCl, 3 mmol/l KCl, 1 mmol/l MgCl[2], 0.1 mmol/l Amplex red [Invitrogen], and 0.35 U/ml horseradish peroxidase [HRP]) in the presence or absence of bovine liver catalase (1500 U/ml). These samples were transferred to 96-well plates and analyzed in a Biotek Synergy 4 microplate reader preheated to 37°C. Amplex Red fluorescence was measured by excitation at 530 nm with a 590-emission filter used to capture emission every min for 1 h and 5 min at 37°C. NOX activity was initiated by the addition of 180 μM NADPH (MP Biomedicals, per well) after the first 5 min of basal activity reading. The mean rate of fluorescence was tabulated by taking the slope (rate) of the linear portion of the response and subtracting the rate of equivalent sample containing catalase (negative control) and averaging the delta of each technical replicates of each biological replicate. The average of the biological replicate plus SEM was reported in all cases. Changes were calculated as fold differences between treatment groups, i.e. Nx vs. Hx. 3.9. Measurements of H[2]O[2] production by coumarin boronic acid (CBA) assay A modified protocol was used as previously described [[87]36]. Briefly, 10 μg of mouse lung lysates or autopsied iPAH patients (per well, as above) were resuspended in CBA assay buffer (25 mM HEPES pH 7.6, containing: 120 mmol/l NaCl, 3 mmol/l KCl, 1 mmol/l MgCl[2], 10 μmol/l DTPA, 100 μmol/l L-NAME and 1 μmol/l taurine ± 1 KU/ml) plus CBA (0.5 mmol/l); ± bovine liver catalase as a negative control. After combination of the lysates with assay buffer plus CBA (± bovine liver catalase), the solution was added to a 96-well plate. Plates were then placed in a Biotek Synergy 4 hybrid multimode microplate reader that was preheated to 37°C and fluorescence read kinetically every min for 4 h at excitation of 350 nm and an emission of 450 nm. The NOX reaction was initiated by the addition of 180 μM NADPH (MP Biomedicals) after the first 5 min of basal activity reading. The average rate of fluorescence was determined as the slope of the linear portion of the response that was subsequently normalized to the slope of the linear portion of the negative control (+ bovine liver catalase) (as above). 3.10. Immunoblotting Western blot was performed as previously described [[88]36]. In brief, lysates of HPAECs or tissue (murine lung/autopsied iPAH human lung) were dissolved 1:1 in 2X Laemmli buffer (Bio-Rad) and heated for 5 min at 95°C. 20–35 μg of proteins were resolved by SDS-page and transferred onto nitrocellulose membrane (Bio-Rad). Nitrocellulose membranes were incubated with Odyssey Blocking Buffer for 1 h at RT (LI-COR Biosciences) and probed with primary antibody overnight at 4°C. The primary antibodies and subsequent dilutions used in this study include: mouse anti-β-actin (Santa Cruz, sc-47778 1:2500), rabbit anti-HIF1α (Abcam, ab179483, 1:500), rabbit anti-NOX1 (Abcam, ab131088, 1:500), rabbit anti-NOXO1 (ThermoFisher, 600-401-899, 1:1000), rabbit anti-NOXA1 (Abclonal, A13844, 1:500), rabbit anti-NOX2 (Abcam, ab80508, 1:500), rabbit anti-NOX4 (Abcam, ab133303, 1:500), goat anti-p22^phox (Santa Cruz, sc-11712, 1:1000), rabbit anti-p47^phox (Millipore Sigma, 07-001, 1:500), goat anti-p67^phox (Santa Cruz, sc-7663, 1:1000), rabbit anti-phospho-eIF2α (Cell Signaling, 9721, 1:500), rabbit anti-eIF2α (Cell Signaling, 5324, 1:500), rabbit anti-BiP (Cell signaling, 3177, 1:1000), rabbit anti-PDI (Cell signaling, 3501, 1:1000), rabbit anti-ATF4 (Cell signaling, 11815, 1:1000), rabbit anti-phospho-PERK (Cell signaling, 3179, 1:1000), rabbit anti-PERK (Cell signaling, 3192, 1:1000). Blots were then probed with anti-rabbit, anti-mouse, or anti-goat secondary antibodies conjugated to fluorophores IRDye 680RD or IRDye 800CW (1:10000 dilution, LI-COR Biosciences). Protein bands were visualized on an Odyssey Imaging System (LI-COR) and their intensity was quantified using ImageJ software. 3.11. Histology sample preparation Lungs were harvested from mice following lung perfusion with 4 % paraformaldehyde (PFA in PBS, ThermoFisher) for preservation of alveoli structures. Thereafter, the tissue was washed in 1X PBS to be devoid of blood and fixed in 4 % PFA overnight at 4°C. On the following day, tissues were washed in 1X PBS and placed in 70 % ethanol at 4°C for 48 h. Pressure-fixed tissues were then paraffin-embedded and sectioned at 2.5–5 μm thickness on glass slides. Before histological- or immuno-staining, tissue sections were de-paraffinized according to previously described procedure [[89]11,[90]36]. 3.12. Hematoxylin & eosin staining After de-paraffinization, sections were washed 2 times in ddH[2]O for 5 min. The sections were submerged in hematoxylin (Lillies Modification) (Abcam, ab245880) and incubated at RT for 10 min. Sections were rinsed twice in ddH[2]O for the removal of excess stain, followed by immersion in “bluing” reagent (Abcam, ab245880) for 15 s at RT. Sections were then rinsed twice in ddH[2]O and immersed in absolute ethanol for 10 s at RT. Sections were submerged in eosin Y solution (Modified Alcoholic) (Abcam, ab245880) and incubated for 5 min at RT. Slides were thereafter dehydrated by immersing them twice in 70 % ethanol for 90 s, followed by immersion twice 96 % ethanol and 100 % ethanol for 90 s each. Finally, sections were immersed twice in xylene for 3 min, thereafter, mounted with Permount (Fisher Chemical), cover-slipped and left to air-dry overnight. Light images were captured on a Nikon A1 microscope. Cross-sectional area (CSA) of the intima (delimited between the lumen and the internal elastic lamina) together with the medial CSA (the area between the internal and external elastic lamina) were quantified via ImageJ. For each slide 4–5 images were captured. Each image's visible field contained at least one randomly selected vessel (40–80 μm diameter). 3.13. Picrosirius red staining After de-paraffinization, sections were washed 2x in ddH[2]O for 5 min. The sections were submerged in Picrosirius red solution (Abcam, ab246832) and incubated at RT for 60 min. Sections were twice rinsed in acetic acid (0.5 %) for the removal of excess stain. Slides were dehydrated twice in 70 % ethanol for 90 s, two times in 96 % ethanol and 100 % ethanol for 90 s each. Finally, sections were immersed twice in xylene for 3 min, thereafter, mounted with Permount (Fisher Chemical) and coverslipped and left to air-dry overnight. Light images were captured on Nikon A1 microscope. Lung vessel collagen deposition/fibrosis was determined by quantification of red staining via ImageJ using color-based (RGB) thresholding as previously described [[91]37,[92]38]. The total area of fibrosis was calculated as a percentage of the total vessel cross-sectional area and was denoted as lung vessel fibrosis (PSR %). For each slide, 4–5 images were captured. Each image's visible field contained at least one randomly selected lung vessel (40–80 μm diameter). 3.14. Immunofluorescent imaging Tissue sections were deparaffinized and for antigen retrieval sections (on slides) were boiled in citric buffer (10 mM Tris, 1 mM EDTA, 0.05 % Tween, 15 min at 90°C, pH 9.0) for 15 min. Slides were washed three times for 5 min in 1x PBS followed by washing three times for 5 min in 1 x PBS +0.2 % Triton X and washed 3 times for 5 min in 1x PBS. Slides were blocked in 1x PBS containing 3 % BSA for 90 min at RT in a humid chamber followed by overnight incubation with primary antibody in the same chamber at 4°C. Primary antibodies were applied in the following dilutions: anti-NOX1 (Abcam, ab131088, 1:200), anti-8-OhdG (Santa-Cruz, sc-66036, pre-conjugated with FITC probe, 1:200) or anti-CD31 aka PECAM-1 (R and D Systems, AF3628, 1:100). After overnight incubation, slides were washed 3 times for 5 min in 1x PBS followed by three washes for 5 min in 1x PBS containing 0.2 % Triton X and washed 3 times for 5 min 1x PBS. Slides were then incubated with Cy3 conjugate or Cy5 conjugate secondary antibodies for 90 min at RT (ThermoFisher Inc., 1:1000). Sections were washed 3 times for 2 min in 1x PBS followed by subsequent staining for nuclei with Hoechst (1mg/100 ml) dye for 1 min at RT in a humid chamber. Lastly, slides were washed 3 times for 2 min in 1x PBS, mounted with gelvatol mounting media (polyvinyl alcohol, glycerol, H[2]O, sodium azide and Tris pH 8.5) and set with coverslips. Non-specific rabbit and goat IgG (5 μg/ml) were used in place of primary antibody as a negative control. Confocal images were captured on Nikon A1 spectral confocal microscope (Nikon Instruments Inc). EC NOX1 and EC 8-OHdG expression were quantified as the average mean intensity of NOX1 or 8-OHdG signal (fluorescence), respectively, where colocalized with CD31 via NIS-Elements Software (Nikon Instruments Inc.) and normalized by subtracting out the signal of negative controls, respectively. CD31^+ endothelial cells were quantified per vessel as the total CD31^+ pixels divided by total vessel pixels (total vessel area), or CD31^+ area under curve (AUC) and represented as a fold change. For each slide, 4–5 images were captured. Each image's visible field contained at least one randomly selected lung vessel (40–80 μm diameter). 3.15. RNA sequencing and data analysis HPAECs were transfected with scrambled (Scr) or NOX1 (siNOX1) siRNA for 48 h before cells were exposed to Nx or Hx for 24 h. At the end of treatment, cells were lysed for RNA isolation and purification (RNeasy MinElute cleanup kit, QIAGEN). Total RNA samples were sent to NovoGene (Sacramento, CA) for mRNA library preparation and sequencing. Unstranded and paired 150 bp sequencing data was filtered, trimmed, aligned to the human reference genome (GRCh38), and counted on the gene level. After filtering low-count genes, the most variable genes (2000 genes) were used for principal component analysis (PCA) using base R functions. Differentially expressed genes (DEGs) were identified using DESeq2 (adjusted p-value <0.05 unless specified otherwise). Pathway enrichment analysis was performed using ClusterProfiler and Gene Ontology biological processes. Pathways with false detection rates (FDR) < 0.05 were considered significant, and redundant pathways (≥50 % overlapping) were removed using the simplify function in ClusterProfiler. 3.16. Wound scratch assay In vitro wound scratch assays were performed as previously described to measure cellular proliferation/migration [[93]36]. In brief, HPAECs were seeded on 6-well plates to achieve 70 % confluence and silenced for ATF4 using siATF4 (ThermoFisher, Assay ID: 122168 [UniGene ID: [94]Hs496487]) or transfected with scrambled RNA (Scr). Alternatively, genetically unaltered cells were treated with integrated stress response inhibitor (ISRIB, 200 nM), a small molecule inhibitor that reverses the effect of activated eIF2α (phospho-eIF2α); its vehicle, DMSO, was used as control (Ctr). In both cases, cells were cultured at 37°C for 48 h. Thereafter, a transversal scratch was made across the well diameter of confluent HPAEC monolayers with a sterile p1000 μl pipette tip and washed with 1x PBS to remove scraped-cell debris followed by a media change. For quantification, three bright-field images were captured at time zero (T0) via a Zeiss Axiovert 40CFL microscope and cells were incubated at 37°C for 24 h in Nx/Hx. At the end of 24 h, three bright field images were captured in the identical position and designated as T24; the percentage wound area closure ([wound area 0 h – would area 24 h/wound area 0 h] X 100) was measured using Image J software. 3.17. Crystal violet proliferation assay In vitro crystal violet assay was performed as previously described to measure cellular proliferation [[95]13]. In brief, HPAECs were seeded in 96-well plates to achieve 60 % confluence and silenced for ATF4 using siATF4 (ThermoFisher, Assay ID: 122168) or transfected with scrambled RNA (Scr). Alternatively, genetically unaltered cells were treated with ISRIB (200 nM) or DMSO (Ctr). In both cases, cells were cultured at 37°C for 24 h. HPAECs were incubated at 37°C for 24 h under Nx/Hx as described above. Cells were washed twice in PBS for 5 min. Sections were stained with 0.5 % crystal violet for 30 min at RT on a bench-top rocker. Subsequently, cells were washed seven times in PBS for 5 min and allowed to dry for at least 2 h. Cells were incubated with 200 μl methanol for 10 min on a rocker for cell solubilization. Absorbance was measured at 595 nm and presented as fold change compared to controls. 3.18. Human specimen collection Human tissue samples were kindly provided by the University of Pittsburgh Pulmonary, Allergy and Critical Care Medicine (PACCSM), Thoracic Tissue Repository (TTR) as blinded and de-identified with no access to private patient information aside from relevant demographic and clinical data found in [96]Supplemental Table 1. Research protocols were approved by the institutional review board (IRB) of the human research protection office (HRPO) of the University of Pittsburgh. IRB documentation was in agreement with the principles outlined in the Declaration of Helsinki, and all subjects gave written informed consent. Strengthening the reporting of observational studies an epidemiology (STROBE) checklist can be found in [97]Supplemental Table 2. Specimens included age and gender-matched lung samples obtained from patients diagnosed with iPAH idiopathic pulmonary arterial hypertension (iPAH, n = 3) and normal lung samples (CTR, n = 3) obtained from organ donors. Tissues were harvested and homogenized on ice in a glass-on-glass mortar and pestle in 200 μl ice-cold lysis buffer in preparation of AR, CBA, or WB (as above). 3.19. Statistical analyses Data are expressed as mean ± SEM. For comparison of results between two data sets, an unpaired Students’ t-test was performed. One-way and two-way ANOVA followed by a Tukey multiple comparison test was used for comparison of results among more than two groups using GraphPad Prism software (version 9). Comparisons with a p-value of 0 ≤ 0.05 were considered statistically significant. 4. Results 4.1. NOX1 induces right ventricle dysfunction and PAH in a pre-clinical model To investigate the in vivo role of NOX1 in the development of pulmonary arterial hypertension (PAH), the mouse Sugen/hypoxia (Su/Hx) model was used. C57BL/6J mice (7–9 per group) were exposed to hypoxia (Hx; 10 % O[2]) for 3 wks following injection with Sugen 5416 (Su, 20 mg/kg; Tocris) on days 0, 7 and 14 ([98]Fig. 1A). Control mice were maintained under normoxic conditions (Nx; 21 % O[2]) and received vehicle injections in place of Sugen ([99]Fig. 1A). Thus, all Hx mice are hereafter referred to as Su/Hx. One day prior to hypoxia exposure (day −1; D[-1]), mice were implanted with osmotic mini pumps to deliver 20 mg/kg/day of either scrambled control peptide (Scr) or the NOX1 specific peptidic inhibitor NOXA1ds for the duration of the study ([100]Fig. 1A). Su was administered under Hx as previously described to promote vascular remodeling [[101]34,[102]39,[103]40]. Fig. 1. [104]Fig. 1 [105]Open in a new tab Selective inhibition of NOX1 ameliorates right ventricle hemodynamic parameters in experimental model of pulmonary arterial hypertension. (A–N) C57BL/6J (WT) mice were intraperitoneally (i.p.) implanted with osmotic pumps delivering 20 mg/kg/day of either scrambled sequence (Scr, control) or specific NOX1 inhibitor (NOXA1ds). Mice were exposed to either hypoxia (Hx, 10 % O[2]) or normoxia (Nx, 21 % O[2]) for 3 wks. Hx mice received an injection of Sugen 5416 (20 mg/kg/day, IP) at days 0, 7, 14 of Hx exposure whereas Nx mice received vehicle (A). (B–H) Values for right ventricle (RV) hemodynamic parameters: representative pressure-volume (P/V) loops (B) and individual value plots for the RV max pressure (RVMP) (C), mean pulmonary artery pressure (mPAP) (D), pulmonary vascular resistance (PVR Woods) (E), maximal rate of RV pressure rise during systole (contraction, dP/dt max) (F), maximal rate of RV pressure decline during diastole (relaxation, dP/dt min) (G), and stroke work (H). Values for the left ventricle (LV) hemodynamic parameters (I–N): representative pressure-volume (P/V) loops (I), and individual value plots for the LV max pressure (LVMP) (J), contractile index (LVCI) (K), maximal rate of LV pressure rise during systole (contraction, dP/dt max) (L), maximal rate of RV pressure decline during diastole (relaxation, dP/dt min) (M) and aortic mean pressure (AOMP) (N). All data were analyzed using a two-way ANOVA followed by a posthoc Tukey's test (C–H, J–N: n = 7–9 mice; ns = non-significant; ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001∗ vs. Nx/Scr; #p < 0.05, ##p < 0.01 vs. Su/Hx). Sugen/hypoxia (Su/Hx) significantly increased right ventricular (RV) max pressure from (RVMP) ∼25–∼55 mmHg ([106]Fig. 1B/C) concomitant with an increase in mean pulmonary arterial pressur e (mPAP) ([107]Fig. 1D) and pulmonary vascular resistance (PVR) ([108]Fig. 1E). NOXA1ds infusion significantly attenuated the Su/Hx-induced increase in RV changes in RVMP and mPAP ([109]Fig. 1C/D) whereas no change was observed in PVR ([110]Fig. 1E). In terms of RV systolic function, there was a ∼55 % increase in the maximum rate of change in RV systolic pressure (RV Max dP/dt) in the Su/Hx group compared to Nx control; an effect that was partially (∼50 %) reversed with NOXA1ds and indicative of a marked amelioration of RV contractility ([111]Fig. 1F). Moreover, Su/Hx caused a ∼77 % reduction in diastolic relaxation, the minimum rate of change in the RV pressure (RV min dP/dt), a response that was partially reversed by NOXA1ds ([112]Fig. 1G). Su/Hx caused a near doubling in stroke work compared to Nx control, an effect nearly abolished by NOXA1ds consistent with improved RV function ([113]Fig. 1H). Neither Su/Hx nor NOXA1ds produced a significant change in left ventricle (LV) hemodynamic parameters (LVMP, LVCI, LV Max/Min dP/dt) and aortic mean pressure (AOMP) ([114]Fig. 1I–N). 4.2. NOX1 induced pulmonary vascular remodeling and lung fibrosis in PAH mice Formalin-fixed paraffin-embedded (FFPE) lung sections were stained with hematoxylin/eosin (H&E) for assessment of lung and pulmonary vasculature remodeling. In support of the PAH phenotype, increased vascular remodeling was detected in pulmonary vessels of approximately 50 μm mean (40–80 μm range) diameter from mice exposed to Su/Hx, as is typically reported for Su/Hx mice [[115]41,[116]42]. ([117]Fig. 2A/B). Specifically, cross-sectional area (CSA) of the vessel, taken as a measure of total vessel wall thickness [[118]41,[119]42], was increased by ∼87.3 % in arteries from Su/Hx vs. Nx mice ([120]Fig. 2A/B). Furthermore, when infused with NOXA1ds, the total vessel wall thickness was attenuated approximately 90 % back to baseline compared to Su/Hx ([121]Fig. 2A/B). Hallmark lung vessel fibrosis [[122]38,[123]43,[124]44], as quantified from picrosirius red staining of vessel collagen deposition [Lung Vessel Fibrosis (PSR%)], was revealed in Su/Hx mice which was abrogated by NOXA1ds ([125]Fig. 2C/D). Fig. 2. [126]Fig. 2 [127]Open in a new tab Selective inhibition of NOX1 attenuates pulmonary vascular remodeling and fibrosis following chronic hypoxic exposure. (A–D) Mice were treated as in experiments described in [128]Fig. 1. (A/B) Formalin-fixed paraffin embedded (FFPE) lung sections were stained with hematoxylin & eosin (H&E) for the assessment of histomorphology. As a measure of pulmonary vascular remodeling (PVR), medial cross-sectional area (CSA) plus neointima CSA, or total vessel wall thickening (% change) was measured in at least five visible fields containing at least one randomly selected vessel (40–80 μm diameter) per lung section (A/B). FFPE lung sections were stained with Picrosirius Red (PSR) for the assessment of pulmonary fibrosis (C/D). Lung vessel collagen deposition and, thus, fibrosis was determined using color-based thresholding. The total area of lung vessel fibrosis was calculated as a percentage of the total vessel cross-sectional area and was denoted as a change in “Lung vessel Fibrosis (PSR%)” from Nx control. For each section, 5 visible fields were captured with at least one randomly selected lung vessel (40–80 μm diameter) (C/D). Representative figures are shown wherein ∗ indicates a bar length of 50 μm while # indicates a bar length of 200 μm. All data were analyzed by two-way ANOVA followed by Tukey's test (A–D: n = 4mice: ∗p < 0.05, ∗∗∗∗p < 0.0001; vs. Nx/Scr; #p < 0.05, ###p < 0.001 vs. Su/Hx). 4.3. NOX1 is activated following sugen/hypoxia (su/Hx) Lungs tissues of Su/Hx mice showed, as expected, an upregulation in hypoxia inducible factor 1 alpha (HIF1α) indicative of hypoxic response, which was unchanged with NOXA1ds infusion (N1ds) ([129]Fig. 3A/B). Moreover, NADPH oxidase (NOX)-dependent H[2]O[2] levels (measured via Amplex Red [AR] or Coumarin Boronate Acid Assay [CBA]) were increased in homogenates of lungs from Su/Hx mice and abolished in Su/Hx mice infused with NOX1 inhibitor ([130]Fig. 3C/D; kinetic H[2]O[2]curves found in [131]Supplemental Figs. 1A–H). This was consistent with an increased “footprint” detection of 8-hydroxy-2-deoxyguanosine (8-OHdG), a surrogate marker of oxidative damage, in the endothelium of lung vessels of Su/Hx mice (measured as the average intensity 8-OHdG cross-reactivity [fluorescence] in areas with colocalization with CD31 fluorescence) ([132]Fig. 3E/F). The 8-OHdG signal in ECs was almost completely prevented in mice infused with NOXA1ds ([133]Fig. 3E/F). Furthermore, an increased NOX1 cross-reactivity was detected in the endothelium of lungs vessels of Su/Hx mice and not changed with infusion of the inhibitor (measured as average intensity of NOX1 fluorescence overlapping with CD31) ([134]Fig. 3E/G). Lastly, total vessel intimal endothelial cells (quantified as CD31^+ pixels/total vessel pixels) increased in Su/Hx mice, with a trend towards partial prevention (p = 0.0598) in Su/Hx mice treated with NOXA1ds ([135]Supplemental Fig. 2A/B). Consistent with histology, lung lysate NOX1 was increased with Su/Hx on a Western blot and not changed with the infusion of inhibitor ([136]Fig. 3H/I). NOX1 activator and organizer subunits (NOXA1 & NOXO1) were upregulated in Su/Hx group ([137]Fig. 3H, J/K). With the infusion of NOXA1ds, NOXO1 was significantly attenuated compared to Su/Hx control littermates, whereas NOXA1 exhibited a tendency toward a decrease ([138]Fig. 3H, J/K). p22^phox levels were upregulated in Su/Hx group but were not changed with the infusion of NOXA1ds ([139]Fig. 3H/L). Fig. 3. [140]Fig. 3 [141]Open in a new tab Hypoxia upregulates and activates canonical NOX1 oxidase in lung endothelium. (A–L) Mice were treated as described in [142]Fig. 1. Western blot analyses were performed on lung (A/B) tissue lysates using monoclonal HIF1α antibody. β-actin served as loading control. Representative blots are shown. (C/D) Freshly explanted and homogenized lung tissues were used for the detection of H[2]O[2] by Amplex Red (C) and coumarin boronate (D) assays. (E–G) Formalin-fixed, paraffin-embedded lung tissue sections were incubated with antibodies against 8-hydroxy-2-deoxyguanosine (8-OHdG) (E/F), NOX1 (E/G), or CD31 (E) and visualized with a secondary antibody conjugated with Alexa Fluor 488, 560–590, or 647. Nuclei were visualized with DAPI. Fluorescence intensity was measured in five visible fields with at least one randomly selected lung vessel (40–80 μm diameter) per lung section. EC NOX1 and EC 8-OHdG expression were quantified as the average mean intensity of NOX1-or 8-OHdG-associated fluorescence, respectively, where colocalized with CD31-associated fluorescence (E–G). Representative images are shown wherein ∗ indicates a bar length of 50 μm. (H–L) Western blot of lung tissue lysates show NOX1 (H/I), NOXO1 (H/J), NOXA1 (H/K), and p22^phox(H/L). β-actin served as loading control. NOXA1ds = N1ds. Representative blots are shown. All data were analyzed by two-way ANOVA followed by Tukey's test (A-D, H–L: n = 7–9 mice, E–G: n = 4 mice; ns = non-significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 vs. Nx/Scr; #p < 0.05, ##p < 0.01, ####p < 0.0001 vs. Su/Hx). Examining other murine vascular NOXs, we observed no change was observed in NOX2 and its cytosolic regulatory subunits, p47^phox and p67^phox ([143]Supplemental Figs. 3A–D). There was an ∼50 % trend toward non-significant upregulation (p = 0.0767) in NOX4 with Su/Hx that was not changed with NOX1 inhibition ([144]Supplemental Fig. 3E/F). 4.4. Unbiased RNA-Seq profiling comparing normoxic and hypoxic NOX1-depleted transcriptome reveals the unfolded protein response To gain insight into how NOX1 might act in the context of Hx, exploratory RNA-Seq was performed on human pulmonary artery endothelial cells (HPAECs) in vitro ([145]Fig. 4). Briefly, HPAECs were NOX1-silenced with NOX1 siRNA (siNOX1) or were transfected with scrambled RNA (control, Scr) and exposed to hypoxia (Hx, 1 % O[2]) or normoxia (Nx, 21 % O[2]) for 24 h. Cells were harvested and RNA-Seq was performed on RNA isolates. In the PCA plot, gene expression changes due to Hx and NOX1 silencing treatment are depicted by changes along the PC1 and PC2 axes, respectively ([146]Fig. 4A). The comparison between Hx/Scr and Nx/Scr in scrambled siRNA-transfected cells showed 2019 significantly (adjusted p-value < 0.05) upregulated genes and 1609 downregulated genes ([147]Fig. 4B). Many of these genes were related categorically to apoptosis, autophagy, cell proliferation, and response to oxygen level changes. Specifically, the following Gene Ontology (GO) biological processes were significantly more common (in order of degree of significance) in these DEGs when comparing Hx/Scr vs. Nx/Scr: extrinsic apoptotic signaling pathway, tRNA metabolic process, stem cell population maintenance, regulation of macroautophagy, cellular response to oxygen levels, negative regulation of mitotic cell cycle, response to topologically incorrect protein, transforming growth factor beta receptor signaling pathway, regulation of intracellular protein transport, protein stabilization ([148]Fig. 4C/D). Notably, a subset of genes was involved in the cellular response to incorrect protein and protein stabilization ([149]Fig. 4C/D). By including an interaction term in the statistical model, we discovered DEGs both induced by Hx and modified by NOX1 silencing ([150]Fig. 4E, p-value < 0.05, 1214 up-regulated, 1517 down-regulated). The genes showing interacting effects of Hx and NOX1 silencing were enriched in similar pathways to those altered by Hx alone ([151]Fig. 4F/G). Interestingly, the interaction showed the unfolded protein response (response to topologically incorrect protein [GO:0035966]) to be one of the pathways that was identified to be induced by Hx and modified by NOX1 silencing ([152]Fig. 4F/G). Specifically, the following GO biological processes were significantly more in common: (in order of degree of significance) G1/S transition of mitotic cell cycle, regulation of cell cycle G1/S phase transition, negative regulation of apoptotic signaling pathway, response to topologically incorrect protein, vitamin metabolic process, cellular response to oxygen levels, cellular response to decreases in oxygen levels, positive regulation of protein catabolic processes, chaperone co-factor dependent protein re folding, and sister chromatid segregation ([153]Fig. 4F/G). Fig. 4. [154]Fig. 4 [155]Open in a new tab RNA sequencing shows NOX1 interacts with hypoxia treatment on genes involved in the unfolded protein response (A–G). Human pulmonary artery endothelial cells (HPAECs) were either silenced for NOX1 (siNOX1) using RNAi or were transfected with scrambled RNA (Scr) and exposed to hypoxia (Hx, 1 % oxygen) or normoxia (Nx, 21 % oxygen) for 24 h. Thereafter, total RNA was isolated, and RNA-Seq transcriptomics profiling was performed. (A) The top 2000 most variable genes, ranked by standard deviations, were used for principal component analysis (PCA). (B) Differential gene analysis (DEG) identified a large number of genes significantly (adjusted p-value [padj]< 0.05) altered by hypoxia treatment alone (2019 up, 1609 down, Hx/Scr vs Nx/Scr). (C) The differentially expressed genes (DEGs, adjusted p-value < 0.05) between Hx/Scr and Nx/Scr were used in an enrichment analysis with Gene Ontology biological processes, and the top 10 pathways ranked by false detection rate (FDR) were included in the dot plot. (D) The chord diagram shows the expression levels of genes (top 100 ranked by padj) involved in these pathways. The color bar indicates gene expression levels in log2 fold changes (Log2FC). (C) and (D) share the same color annotations and order for the pathways. (E) Genes showing significant interaction effects (p-value < 0.05) of Hx and NOX1 silencing were highlighted in the volcano plot (1214 up, 1517 down). The top 10 significantly enriched Gene Ontology biological processes and the corresponding gene expressions were shown in a dot plot (F) and a chord diagram (G), which share the same pathway order and color annotations. The color scale for interaction in (G) represents the log2FC of Hx/siNOX1 over the additive effects of Hx/Scr and Nx/siNOX1. (A–G: n = 3 biological replicates of cells) Of the GO biological processes that were significantly enriched, we investigated the “response to topologically incorrect protein” process (GO:0035966) [containing proteins involved in the unfolded protein response (UPR)]. We interrogated genes that were elevated in Hx/Scr (vs. Nx/Scr) and then downregulated with siNOX1 ([156]Fig. 4C/D, F/G). Moreover, a previous report broadly implicated proteins in the UPR pathway PERK in a model of PAH [[157]45]. Thus, the PERK arm of UPR, per se, became the primary focus of our analysis. As such, we examined a comprehensive array of common proximal and distal genes of the PERK pathway and found a majority to be transcriptionally upregulated in response to Hx. Genes that did not increase in transcription in response to hypoxia, or were not implicated to be associated with proliferation in the literature [[158]46], were excluded. Intriguingly, most were partially if not fully prevented when treated with NOX1 siRNA i.e. PDIA3, PDIA4, PDIA6, eIF2β2, elF2AK3, GDF15, PTGS2, S100A6, NFE2L3, LAMP1, CDH24, and CDCP1 ([159]Supplemental Figs. 4A–M). 4.5. NOX1-derived ROS induce the unfolded protein response in vivo To explore activation of UPR pathways in PAH mice, Western blot was conducted on murine lung lysates ([160]Fig. 5). Indeed, increased phosphorylation of PERK as a measure of its activity (phospho-vs. total PERK) was detected in lysates of Su/Hx vs. Nx mice (∼75 % vs. Nx) and the p-PERK/t-PERK increase was abolished by NOXA1ds (N1ds) ([161]Fig. 5A/D). Similarly, phosphorylation and activation of eIF2α was markedly increased (∼2–3 fold; [162]Fig. 5B/E) and negated by NOXA1ds. ATF4, PDI, and binding immunoglobin protein (BiP) followed the same pattern ([163]Fig. 5C & F-H). Fig. 5. [164]Fig. 5 [165]Open in a new tab NOX1-derived ROS induce the unfolded protein response in murine lungs following hypoxic exposure. (A–H) Lung lysates of mice (treated as in experiments described for [166]Fig. 1, [167]Fig. 2, [168]Fig. 3) were evaluated by Western blot using antibodies against: phospho- and total PERK (A/D), phospho- and total eIF2α (B/E), ATF4 (C/F), PDI (C/G), and BiP (C/H). β-actin or total protein levels (t-PERK or t-eIF2α) served as loading control or for activity comparison to total protein amounts. NOXA1ds = N1ds. Representative blots are shown. All data were analyzed using a two-way ANOVA followed by Tukey's test (A–H: n = 7–9 mice); ∗p < 0.05, ∗∗p < 0.01 vs. Scr/Nx; #p < 0.05, ###p < 0.001 vs. Su/Hx). N1ds = NOXA1ds. 4.6. NOX1 induces the unfolded protein response in HPAECs in vitro Next, we interrogated the role of NOX1 in UPR and cellular proliferation and migration. HPAECs were treated with NOXA1ds (N1ds) or scrambled peptide (Scr) and exposed to either Hx or Nx for 24 h. Hypoxic HPAECs displayed elevated HIF1α protein levels ([169]Fig. 6A/D) concomitant with PERK ([170]Fig. 6B/E) and eIF2α ([171]Fig. 6C/F) activation with near complete ablation of this elevation of activated PERK and eIF2α by NOXA1ds. In line with these findings, we observed hypoxic activation of the PERK/eIF2a arm of the UPR pathway given that downstream mediators ATF4 ([172]Fig. 6G/H), PDI ([173]Fig. 6G/I) and BiP ([174]Fig. 6G/J) followed the same pattern. Fig. 6. [175]Fig. 6 [176]Open in a new tab NOX1-derived ROS induce the unfolded protein response in HPAECs in vitro. (A–J) Human pulmonary artery endothelial cells (HPAECs) were treated with NOX1 inhibitor (NOXA1ds, N1ds) or control peptide (Scr) and exposed to hypoxia (Hx, 1 % oxygen) or normoxia (Nx, 21 % oxygen) for 24 h. Western blot analyses were performed on cell lysates using antibodies against: HIF1α (A/D), phospho- and total PERK (B/E), phospho- and total eIF2α (C/F), ATF4 (G/H), PDI (G/I), and BiP (G/J). β-actin or total protein levels (t-PERK/t-eIF2α) served as loading control or for activity comparison to total protein amounts. Representative blots are shown. (K/L) HPAECs were seeded in 6-well plates and either silenced for ATF4 using RNAi or transfected with Scr RNA. Thereafter at 100 % confluency, a void was made across the well and demarcated and HPAECS were exposed to either Hx or Nx for 24 h (K/L). Alternatively, HPAECs treated with either with a selective inhibitor of eIF2α dephosphorylation (ISRIB, 200 nM) or vehicle control (DMSO, Ctr) exposed to either Hx or Nx for 24 h (N/O). Photos show cellular front line after 24 h compared with starting cellular front (black line) (K/L, N/O). Representative images are shown wherein ∗ indicates a bar length of 100 μm (K/L). Data are shown as relative gap size reduction (wound closure) after 24 h (L/O) Alternatively, HPAECs were silenced for ATF4 (M) or inhibited with eIF2α (P) (as above) and exposed to Hx or Nx for 24 h. Thereafter, HPAECs were assessed for proliferation via crystal violet assay (M/P). All data were analyzed with two-way ANOVA followed by Tukey's test (A–P: n = 7–10 biological replicates of cells; ns = non-significant; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001 vs. Nx/Scr; #p < 0.05, ##p < 0.01, ###p < 0.001, ####p < 0.0001 vs. Hx/Scr). Causality of ATF4 and eIF2α with respect to EC cellular proliferation and migration was tested by examining the effect of blockade of ATF4 and eIF2α. As such, these processes were interrogated using HPAECs transfected with either siATF4 vs. scrambled RNA (Scr) ([177]Fig. 6K–M, [178]Supplemental Fig. 5A/C/E/H) or phospho-eIF2α antagonist – integrated stress response inhibitor (ISRIB) vs. vehicle DMSO (Ctr) ([179]Fig. 6N–P, [180]Supplemental Fig. 5B/D/F/I). In both cases, cells were exposed to Hx or Nx for 24 h and the wound healing scratch assay ([181]Fig. 6K/L and [182]Fig. 6N/O) or crystal violet proliferation assay ([183]Fig. 6M/P, [184]Supplemental Figs. 5A–J) were employed to assess effects on cellular proliferation/migration or proliferation, respectively. Following Hx, the extent of cellular proliferation/migration (wound closure) and proliferation (crystal violet) increased significantly and consistent with NOX1-mediated ATF4 and elF2α regulating these processes, both were abrogated with either siATF4 or ISRIB, respectively ([185]Fig. 6K–P, [186]Supplemental Figs. 5A–J). Growth curves of cells treated as above confirmed proliferation results at various timepoints (0h, 4h, 8h, 12h, 24h) ([187]Supplemental Figs. 5A–J). 4.7. NOX1-overexpression and induction of the unfolded protein response in human iPAH To probe clinical significance in human patients, lung lysates from human subjects diagnosed with idiopathic pulmonary arterial hypertension (iPAH) were analyzed and compared with control subjects for the expression of UPR markers. Demographic data are presented in [188]Supplemental Table 1. Both HIF1α and NOX1 were markedly increased in iPAH patient lungs compared to control subjects ([189]Fig. 7A/B, E/F). Furthermore, H[2]O[2] was increased or trended toward (p = 0.0657 vs Ctr) an increase with iPAH as detected by Amplex Red or CBA, respectively ([190]Fig. 7C/D; kinetic ROS [H[2]O[2]] curves found in [191]Supplemental Figs. 6A–D). Fig. 7. [192]Fig. 7 [193]Open in a new tab Human iPAH lungs exhibit overexpression of NOX1 and connection with the unfolded protein response. (A/B, E-N) Western blot of iPAH or CTR lungs was conducted using antibodies against: HIF1α (A/B), NOX1 (E/F), phospho- and total PERK (G/H), phospho- and total eIF2α (I/J), ATF4 (K/L), PDI (K/M), and BiP (K/N). β-actin or total protein levels (t-PERK or t-eIF2α) served as control and representative blots are shown. (C/D) Amplex Red (C) and coumarin boronate (D) assays were employed for H[2]O[2] detection. All data were analyzed with a two-way ANOVA followed by Tukey's test (A–N: n = 3 biological replicates, human lung samples; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, vs. Ctr. For Amplex Red measurement of H[2]O[2] comparisons p = 0.0657 vs. Ctr; for BiP comparisons p = 0.0891 vs. Ctr.). Paralleling the observations in mice and hypoxic HPAECs, phosphorylation/activation of PERK ([194]Fig. 7G/H) as well as eIF2α ([195]Fig. 7I/J), ATF4 ([196]Fig. 7K/L) and PDI ([197]Fig. 7K/M) were significantly increased in tissue from iPAH as compared to control subjects. BiP levels showed a numerical increase (p = 0.0891 vs Ctr) that did not reach statistical significance ([198]Fig. 7K/N). Taken together, these results substantiate increased NOX1 and promotion of the PERK/eIF2α/ATF4 axis of the UPR in human pathology. 5. Discussion We report here for the first time the role of NOX1 in mediating the signaling pivotal to the unfolded protein response (UPR) in hypoxic human pulmonary artery endothelial cells (HPAECs) in vitro. Clinical samples of patients with iPAH and NOX1-inhibited Sugen/Hypoxia (Su/Hx) mice supported the notion of NOX1 being central to the disease development and progression. Indeed, the increased NOX1 expression correlated with disease in iPAH patients, whereas with mice, specific NOX1 inhibition, acted as partially preventive of PAH development. Mechanistically in hypoxic HPAECs, NOX1-dependent proliferative and migratory responses are significant contributors to pulmonary vascular remodeling in PAH [[199]15,[200]16]. In an unbiased search employing RNA-Seq on human pulmonary artery ECs, we identified the unfolded protein response as a pathway that is upregulated by hypoxia and reversed by knockdown of NOX1. Indeed, empirical data from Su/Hx mouse lungs, hypoxic human pulmonary ECs, and human iPAH lungs revealed that the PERK arm of UPR pathways is induced, at least in part, by NOX1 in PAH and is clinically significant. Moreover, we were also able to show that NOX1-mediated PERK, and by extension UPR, is causally linked to lung EC proliferation and migration. In aggregate, our findings in human ECs in vitro and in clinical samples and NOX1-inhibited Su/Hx mice are in accord with NOX1 as pivotal to a unique and novel UPR activation participating in the development of PAH. With respect to the pathological presentation of the disease, we report that inhibition of NOX1 elicits ameliorative effects on hemodynamic parameters in Sugen5416 + hypoxia (Su/Hx) mice manifested in mitigation of pulmonary vascular remodeling and improvement in RV dysfunction. Importantly, Su/Hx, an in vivo model of PAH, employed in this study recapitulated early-stages of PAH reflected by an increased vessel occlusion due to vascular wall thickening [[201][11], [202][12], [203][13],[204]34,[205]35,[206]39,[207]40,[208]47] concomitant with an increase in mPAP, RV pressure and contractility mirrored by an RV adaptation to pressure overload through overall workload increase [[209]48,[210]49]. Moreover, our studies herein of HPAEC proliferation in response to hypoxia employ a model of choice in vitro in line with PAH [[211][11], [212][12], [213][13], [214][50]]. A potential limitation of this model, however, is that while it represents and accurately models characteristics of Group I (PAH), it may also mimic certain aspects of Group III PH that are associated with lung disease and/or hypoxia. As such, a deeper examination of this implication is warranted in future studies. Su/Hx elicited characteristic alterations in RV and pulmonary artery pressure, PVR, RV contraction and relaxation, RV stroke work that were partially prevented by NOX1 inhibitor (NOXA1ds). In contrast, neither Hx nor NOX1 inhibitor demonstrated an effect on the LV. The latter demonstrates the relative specificity of Su/Hx to alter pulmonary hemodynamics and the right heart, per se [[215][39], [216][40], [217][41]]. Importantly, NOX1 inhibitor, NOXA1ds partially reversed most of right heart-related perturbations with the exception of PVR Woods measurements. That is, although unexpected, we did not observe a significant change in PVR upon treatment with NOXA1ds. This unexpected finding in our hands may have arisen as a limitation of conventional tabulation of PVR (aka PVR Woods), i.e. PVR Woods equals mPAP minus pulmonary arterial wedge pressure (PAWP)/RV cardiac output X 80. In other words, while PVR was characteristically upregulated in Su/Hx, it did not appear to be attenuated with the infusion of NOXA1ds. This seemingly counterintuitive finding may be ascribed to a technical limitation of measuring vanishingly small RV volumes in mice and, thus, the high variability of values we tabulated of RV cardiac output (RVCO) (see [218]Supplemental Fig. 7). This high degree of variability appears to have confounded any differences we might have otherwise observed. Nevertheless, we did observe Su/Hx-induced vascular remodeling characterized by vessel wall thickening [[219]39,[220]41,[221]42] represented by the increased thickening of pulmonary vessels in the Su/Hx group and its reversal with the administration of NOXA1ds. Increased thickening, commonly referred to as muscularization, has been associated with an increased propensity for constriction of pulmonary arteries in PAH [[222]2,[223]3] which has long-justified the use of vasodilators to alleviate pressure (PAP; RV afterload) and extend RV health [[224][2], [225][3], [226][4],[227]51]. Moreover, we found that the intimal layer of the pulmonary arteries was significantly increased by Su/Hx and NoxA1ds markedly attenuated this response. To add to these changes, emblematic elevations in fibrosis might be seen as adding to the rigidity of pulmonary vessels and, therefore, increased resistance and mPAP. Notably, NOXA1ds also markedly attenuated the fibrotic response. Our previous work implicated NOX1 as a driver of BMPR2-and Sp1-mediated EC signaling in mouse and human PAH [[228][11], [229][12], [230][13]] and other studies of pulmonary artery smooth muscle cells (PASMCs) have suggested NOX1 as a mediator in PAH-related remodeling [[231][14], [232][15], [233][16],[234]52]. Despite this, examination of the pathways instigated by NOX1 in rodent and human PAH are underexplored, no less the therapeutic value of targeting NOX1 in PAH in vivo. In sharp contrast to this notion, Iwata and coworkers [[235]53] reported spontaneous PVR in NOX1-deficient mice which was rescued with a NOX1 transgene [[236]53]. This discrepancy might be attributed to the use in that study of a genetic KO from conception [[237]53], in contraposition to our current study in which NOX1i was administered to adult mice. In this study, NOX1 protein expression was increased in Su/Hx mouse lung tissue (colocalizing with the endothelium), in hypoxic HPAECs in vitro, and lungs of human iPAH, which make the case for a role of NOX1 in remodeling and RV hemodynamic changes. In support of NOX1 induction and activation, in Su/Hx mice the expression of canonical NOX1 oxidase subunits, NOX1, NOXA1, NOXO1, and p22^phox (membrane), was increased concomitant to the increase in H[2]O[2] levels. Only NOXO1 levels significantly decreased from stimulated levels with the infusion of NOXA1ds. While this observation is not necessary to explain the effect of the NOX1 inhibitor, the modulation observed with NOXO1 appears to suggest a feed-forward NOX1/ROS-induced effect on transcription/translation on NOXO1 per se. Further to the effect of NOXA1ds, it is an isoform-selective inhibitor of NOX1 generated to mimic the activation domain of NOXA1 and thus competitively blocks a crucial internal interaction of NOX1 oxidase pivotal to H[2]O[2] generation (by way of superoxide anion, O[2]^.-) [[238]12,[239]13,[240]54]. Importantly, NOXA1ds’ capacity for inhibition of NOX1-derived ROS in vivo is evidenced by the attenuation of stimulated H[2]O[2] levels we detected in Su/Hx lung lysates. As for its phenotypic effects, we observed this capacity of NOXA1ds to abrogate H[2]O[2] to be concomitant with hemodynamic and morphometric improvements. It has been suggested that the ER stress and induction of UPR leads to cellular survival and proliferation in a variety of cell types [[241][17], [242][18], [243][19]]. Hypoxia-induced accumulation of unfolded proteins and activation of the UPR, for instance in SMCs, leads to increased glycolysis, cell viability, and proliferation [[244]45]. Indeed, not surprisingly, a literature search revealed an association between UPR and PAH [[245]45,[246][55], [247][56], [248][57]]. Discoveries of exacerbation of UPR pathway responses in PAH have been more recent. Four-phenylbutyric acid (4-PBA; UPR blocker) was shown to ameliorate the effects of the hypoxia-induced PAH in mice [[249]58] as well as in a model of monocrotaline-induced PAH in rats [[250]57]. Interestingly, Koyama et al. [[251]58] observed activation of the ATF6 pathway in mice subjected to hypoxia in vivo and in cultured vascular SMCs, but, surprisingly, they did not observe activation of the CHOP pathway in their studies [[252]58]. In contrast, Wu et al. [[253]57] reported activation of all three UPR branches in MCT-induced PAH in rats [[254]57]. The same study [[255]57] suggested that endothelial cells (through BCL-2) may be responsible for the elevation of ER stress/UPR. Moreover, a rare report that investigated UPR in ECs in human PH examined expression of UPR markers in lung vasculature of patients with systemic sclerosis-associated PAH and found elevated ER stress markers BiP and CHOP compared to healthy subjects [[256]59]. Those studies indicate that the pathogenesis of PAH is at least correlated with ER stress and UPR in ECs. In agreement with this notion, our unbiased RNA-Seq data identified the UPR signaling network as a major pathway upregulated in ECs in response to hypoxia. Moreover, our data show that UPR pathway enrichment is reversed by siNOX1. As such, a possible regulatory role of NOX1 emerges in the hypoxic induction of the UPR in PAH. On closer scrutiny, a previous study implicating PASMCs in PAH showed the PERK arm to be the key arm of the UPR pathway in a BMPR2 mutant model of PAH and that PERK inhibition was cardioprotective and preventative of PVR [[257]45]. This suggested that NOX1 might regulate PERK and UPR in ECs as well. Indeed, we have identified the PERK arm of the UPR as upregulated by NOX1 in ECs and thus provide a link between NOX1-mediated PERK activation and EC proliferation and migration. Together with PERK activation in iPAH subjects, our in vitro and in vivo data support the concept that NOX1 plays a significant role in UPR activation and development of PAH. As alluded to above, our unbiased search employing RNA-Seq on human pulmonary artery ECs identified the UPR as a major pathway upregulated hypoxia and reversed by knockdown of NOX1. And, molecular characterization of Su/Hx mouse lungs, hypoxic human pulmonary ECs, and human iPAH lungs revealed that the PERK arm of UPR pathways is mediated by NOX1 in PAH and is clinically significant. Uniquely, we were also able to show that NOX1-mediated PERK is causally linked to lung EC proliferation in PAH. A panel of salient proteins of the PERK arm of the pathway were transcriptionally upregulated by hypoxia and mitigated by NOX1i rendering their upregulation NOX1-dependent. These include eIF2β2, redox-sensitive proteins and activators of PERK (PDIA3, PDIA4, PDIA6), and several pro-proliferative ATF4 transcription factors (TFs: GDF15, PTGS2, S100A6, NFE2L3, LAMP1, CDH24, and CDCP1) [[258]46]. Importantly, there were multiple genes that were not altered by hypoxia, and we also found expression of elF2AK3 (PERK) and eIF2α to be increased under hypoxia with no measurable change upon treatment with siNOX1. We are not entirely clear why these particular proteins were not modulated by NOX1 like the others in their class. However, one explanation might be that the transcription of these genes is not regulated by NOX1 but that eIF2α and PERK post-translational activation are. Indeed, PERK and UPR have been associated with ER stress in PAH [[259]45,[260]57,[261]60]. In contrast, our findings delve into the mechanisms driving this pathway and are consistent with a previously undefined role for NOX1-derived ROS in PERK activation and UPR. One possibility is that the activation of PERK in this scenario is controlled by NOX1-mediated elevations in and oxidation of PDI [[262]61], which in turn can oxidize PERK. The former is consistent with the increased PDI levels we observed in the lungs of the Su/Hx group that are attenuated upon NOXA1ds treatment. Alternatively, it is plausible that NOX1-derived-ROS are involved in direct oxidation and activation of PERK as has been shown by other mediators of oxidation in cells [[263]62]. Our findings are congruent with upregulation (and activation in the case of PERK and elF2α) of the PERK/eIF2α/ATF4 branch of the UPR mediated by NOX1 in Su/Hx lung lysates, ECs, and iPAH human tissue. In support of causality, proliferation and migration were also blunted upon blocking key focal points of the pathway. While inhibition of the UPR via (a) inhibition of elF2α [[264]63,[265]64] by employing integrated stress response inhibitor (ISRIB); or (b) suppression of ATF4 [[266]65,[267]66] by RNAi has indeed been shown to decrease cellular proliferation and migration in the endothelium, this has never before been described to our knowledge in ECs in the etiology of PAH or under the control of NOX1 in any scenario for that matter, physiological or pathological. Concordantly, NOX1's regulatory control of elF2α or ATF4 being linked to UPR and EC proliferation has not previously been identified in any milieu. The current data are further substantiated by our previous studies that had showed NOX1 as an instigator of the pro-proliferative cellular signaling through either bone morphogenic protein (BMP) antagonist gremlin1 [[268]11,[269]13] (Grem1) or the chemokine CXCL12 [[270]12]. Specifically, in a report by Al Ghouleh et al., [[271]11] NOX1-derived ROS activated sonic hedgehog protein (SHH), elicited increased Grem1 expression, an antagonist to BMP4 [[272]67] (whose signaling is crucial to maintenance of endothelial homeostasis), and led to enhanced EC proliferation/dysfunction in PAH [[273][68], [274][69], [275][70]]. Intriguingly, ATF4 has been shown to transcriptionally upregulate SHH in gastric cancer cells, ultimately leading to enhanced proliferation, invasion, and migration [[276]66]. Hence, one other mechanism by which upregulation of ATF4 could cause EC proliferation may be through SHH & Grem1 signaling [[277]11]. Herein, we build on those findings [[278]11,[279]13] and propose UPR as a possible “missing link” between NOX1-derived ROS and activation of SHH, leading to the ultimate promotion of EC proliferation in PAH. In fact, the promotion of ATF4 leads to upregulation of SHH/Grem1 and myriad other transcription factors, both canonical and non-canonical to the UPR that might promote EC proliferation and PVR in PAH. In further support of a UPR-driven EC proliferation, a panel of pro-proliferative genes reported downstream of ATF4 (RNA-Seq) alongside SHH have been related to PAH and cellular proliferative responses downstream of the transcription factor ATF4. These include growth differential factor 15 (GDF15), prostaglandin-endoperoxide synthase 2 (PTGS2), S100 calcium binding protein A6 (S100A6), lysosomal associated membrane protein 1 (LAMP1), cadherin 24 (CDH24), and CUB domain containing protein 1 (CDCP1). GDF15 has been shown to be abundantly expressed in human plexiform lesions (found in higher forms of PAH) and pulmonary EC/SMCs of iPAH patients [[280]71], while PTG2S has been shown upregulated in iPAH PSMCs [[281]72]. S100A6 has been shown to cause proliferation, invasion, and migration of lung cancer cells [[282]73], and CDCP1 has been shown to be inducible by hypoxia via HIF2 [[283]74,[284]75]. Taken together, this points to ATF4's promotion of EC proliferation/migration in which multiple pro-proliferative downstream mediators of ATF4 are plausible. Several limitations must be considered in the current study. Firstly, the contribution of SMCs, and more specifically SMC NOX1, to the pathology of PAH was not investigated. It is known that PASMC proliferation [[285]76,[286]77] and migration [[287][78], [288][79], [289][80]] are major contributors to PVR and PAH. However, the primary focus of this study was ECs and the testing of a novel therapeutic approach for PAH management based on NOX1 inhibition. Ongoing studies in the laboratory are delving into the contribution of individual vascular layers along with the effect of EC vs SMC NOX1 in PAH. Secondly, this study did not investigate the effect of NOXA1ds in vivo in females. Given the known impact of sex on cardiovascular pathophysiology [[290]81] and PAH, i.e. higher prevalence in females at child-bearing age [[291][82], [292][83], [293][84]], admittedly this is not a trivial shortcoming. Nevertheless, we chose to focus these foundational studies on male mice owing to the knowledge that Sugen and estrogen interfere with, and potentially neutralize, each other's mode of action [[294][85], [295][86], [296][87], [297][88], [298][89], [299][90], [300][91], [301][92]]. Thirdly, this study does not delineate the temporal relationship between NOX1 activation, ROS formation, activation of the UPR, and eventual PVR/development of PAH. To elucidate this, future studies are required to investigate NOX1, the UPR, and vascular remodeling over the time course of PAH development. In summary, the findings of this study uncover a novel signaling cascade in PAH in which pulmonary vascular endothelial NOX1 is activated, leading to an increase in ROS, activation of PERK-mediated UPR causing EC proliferation/migration, PVR, and ultimate pathogenesis of PAH. These findings are supported by our data collected in vitro (Hx HPAECs/RNA-Seq), in vivo (Su/Hx Mice) and in human iPAH lung tissue, and underscore clinical significance. Further, identification of the NOX1-UPR (PERK) signaling axis reveals potential new therapeutic target(s) in PAH. This, combined with insights into this cascade in human iPAH lung tissue points to proximal inhibition of NOX1 to be clinically relevant in the future treatment of PAH. Finally, these results carry therapeutic value for the treatment of PAH vis-à-vis PERK inhibition. While the integrated stress response inhibitor (ISRIB) has been described for possible in vivo blockade of PERK and UPR, its insolubility renders it unsuitable as a therapy [[302][93], [303][94], [304][95], [305][96]]. Hence, although in it is early days, targeting NOX1 could offer a new opportunity and potential new target in the treatment of PERK, UPR and PAH. CRediT authorship contribution statement Christian J. Goossen: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Alex Kufner: Investigation. Christopher M. Dustin: Writing – review & editing, Investigation. Imad Al Ghouleh: Writing – review & editing, Methodology. Shuai Yuan: Writing – review & editing, Formal analysis, Data curation. Adam C. Straub: Writing – review & editing. John Sembrat: Resources. Jeffrey J. Baust: Writing – review & editing, Methodology, Investigation. Delphine Gomez: Methodology, Data curation. Damir Kračun: Writing – review & editing, Writing – original draft, Validation, Supervision, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Patrick J. Pagano: Writing – review & editing, Writing – original draft, Supervision, Software, Resources, Project administration, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. 1. Translational perspective Pulmonary arterial hypertension (PAH) is a devastating disease without a cure and a poor understanding of its pathognomonic mechanisms is a barrier to clinical advances. Here, we show that NOX1 participates in pulmonary endothelial pro-proliferative signaling involving the unfolded protein response in human cells and tissue from PAH patients. From a clinical standpoint, ameliorative effects of a selective NOX1 inhibitor in rodent PAH hold promise for a novel therapeutic target for the disease. Non-Standard Abbreviations and Acronyms 8-OhdG 8-hydroxy-2-deoxyguanosine AOMP aortic mean pressure ATF4 activating transcription factor 4 ATF6 activating transcription factor 6 BiP binding immunoglobulin protein BMP bone morphogenic protein CDCP1 CUB domain containing protein 1 CDH24 cadherin 24 CXCL12 stromal cell-derived factor 1 DEGs differentially expressed genes ECs endothelial cells eIF2α eukaryotic translation initiation factor 2A eIF2β2 eukaryotic translation initiation factor 2B subunit beta FDR false detection rate FFPE formalin fixed paraffin embedded GDF15 growth/differentiation factor 15 GEO gene expression omnibus Grem1 gremlin 1 H/E hematoxylin & eosin HIF1α hypoxia inducible factor 1 alpha HIF2α hypoxia inducible factor 2 alpha HPAEC human pulmonary artery endothelial cells HRP horseradish peroxidase Hx hypoxia/hypoxic IP intraperitoneally iPAH idiopathic pulmonary arterial hypertension IRE1α inositol-requiring enzyme-1 alpha ISRIB integrated stress response inhibitor LAMP1 lysosomal-associated membrane protein 1 LV left ventricle LVCI left ventricle contractile index LVMP left ventricle max pressure mPAP mean pulmonary artery pressure NFE2L3 NFE2 like BZIP transcription factor 3 NOX1-5 NADPH oxidase 1-5 NOXA1 NADPH oxidase activator 1 NOXA1ds specific peptidic NOX1 inhibitor NOXO1 NADPH oxidase organizer 1 Nx Normoxia/Normoxic p22^phox cytochrome b light chain p40^phox neutrophil cytosolic factor 4 p47^phox neutrophil cytosolic factor 1 p67^phox neutrophil cytosolic factor 2 PAH pulmonary arterial hypertension PDIA3-6 protein disulfide isomerase family A member 3-6 PERK protein kinase R-like endoplasmic reticulum kinase PASMCs pulmonary artery smooth muscle cells PTGS2 prostaglandin-endoperoxide synthase 2 PV loop pressure/volume loop PVR pulmonary vascular remodeling ROS reactive oxygen species RV right ventricle RVH right ventricle hypertrophy RVMP RV max pressure S100A6 S100 calcium binding protein A6 SOD superoxide dismutase Su Sugen5416 UPR unfolded protein response XO xanthine oxidase [306]Open in a new tab Data availability statement The data related to this article are incorporated into the article and its supplementary material. Raw bulk RNA sequencing data have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus ([307]GSE291410). Funding This work was supported by R01 HL142248 (P.J.P.), AHA 18TPA34170069 (P.J.P.), P01 HL103455 (P.J.P.)., R35 HL161177, (A.C.S) T32 GM133332, and F31 1HL167563 (C.J.G.). Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements