PHD2 Deficiency in Endothelial Cells and Hematopoietic Cells Induces Obliterative Vascular Remodeling and Severe Pulmonary Arterial Hypertension in Mice and Humans through HIF-2α

Abstract

Background

Vascular occlusion and complex plexiform lesions are hallmarks of the pathology of severe pulmonary arterial hypertension (PAH) in patients. However, mechanisms of obliterative vascular remodeling remain elusive and hence current therapies have not targeted the fundamental disease modifying mechanisms and result in only modest improvement in morbidity and mortality.

Methods and Results

Mice with Tie2Cre-mediated disruption of Egln1 (encoding prolyl-4 hydroxylase 2, PHD2) (Egln1^Tie2) in endothelial cells (ECs) and hematopoietic cells exhibited spontaneous severe PAH with extensive pulmonary vascular remodeling including vascular occlusion and plexiform-like lesions resembling the hallmarks of the pathology of clinical PAH. As seen in idiopathic PAH patients, Egln1^Tie2 mice exhibited unprecedented right ventricular hypertrophy and failure and progressive mortality. Consistently, PHD2 expression was diminished in lung ECs of obliterated pulmonary vessels in idiopathic PAH patients. Genetic deletions of both Egln1 and Hif1a or Egln1 and Hif2a identified hypoxia-inducible factor-2α (HIF-2α) as the critical mediator of severe PAH seen in Egln1^Tie2 mice. We also observed altered expression of many PH-causing genes in Egln1^Tie2 lungs which was also normalized in Egln1^Tie2/Hif2a^Tie2 lungs. PHD2-deficient ECs promoted smooth muscle cell proliferation in part through HIF-2α-activated CXCL12 expression. Genetic deletion of Cxcl12 attenuated PAH in Egln1^Tie2 mice.

Conclusions

These studies defined an unexpected role of PHD2 deficiency in the mechanisms of severe PAH and identified the first genetically modified mouse model with obliterative vascular remodeling and pathophysiology recapitulating clinical PAH. Thus, targeting PHD2/HIF-2α signaling is a promising strategy to reverse vascular remodeling for treatment of severe PAH.

Introduction

Pulmonary arterial hypertension (PAH) is characterized by progressive increase of pulmonary vascular resistance and obliterative pulmonary vascular remodeling that result in right heart hypertrophy, failure and premature death. ¹ The histopathological features of PAH includes intima and media thickness, muscularization of distal pulmonary arteries, vascular occlusion and complex plexiform lesions. ^2–5 Owing to the poor understanding of the underlying mechanisms of obliterative vascular remodeling, current therapies targeting abnormalities in the prostacyclin, nitric oxide, and endothelin pathways result in only modest improvement in morbidity and mortality. ^6–8 One of the major hurdles is the lack of appropriate murine models recapitulating the pathophysiology of severe PAH in patients including idiopathic PAH (IPAH). ^9,10 Thus, identifying such genetically modified mouse model(s) and elucidating the molecular mechanisms of obliterative pulmonary vascular remodeling will provide valuable druggable targets and novel therapeutic approaches for effective treatment of PAH in patients.

Hypoxia-inducible factors (HIFs) comprised of an O2-sensitive α-subunit (mainly HIF-1α and HIF-2α) and a constitutively expressed β-subunit are key transcription factors mediating adaptive responses to hypoxia and ischemia. ^11,12 As O2 sensors, HIF prolyl-4 hydroxylases [prolyl hydroxylase domain-containing enzymes (PHDs), also known as EGLN1–3] use molecular O2 as a substrate to hydroxylate specific proline residues of HIF-α. Hydroxylation promotes HIF-α binding to the von Hippel-Lindau (VHL) ubiquitin E3 ligase resulting in ubiquitination and subsequent degradation by proteasome. ^13–16 Under hypoxic condition, inhibited PHD activity results in stabilization and accumulation of HIF-α in the nucleus. HIF-α and HIF-β form a heterodimer and activate expression of target genes that regulate angiogenesis, erythropoiesis, metabolism, inflammation, and vascular responses ^{11, 12, 17}. It has been shown that HIF-α activation plays an important role in the pathogenesis of pulmonary hypertension (PH). Hif1a^+/− mice exhibited decreased PH whereas Hif2a^+/− mice fail to develop PH in response to chronic hypoxia.^{18, 19} Inactivation of Hif1a in smooth muscle cells (SMCs) in adult also attenuates hypoxia-induced PH. ²⁰ An activating mutation of Hif2a induces PAH in mice and humans.^21,22 Vhl mutation also induces PH in mice and humans. ^23,24 However, little is known about the role of PHDs in the pathogenesis of PH. Pharmacological and genetic evidence showed that PHD2 is responsible for the majority of HIF-α hydroxylation while PHD1 and PHD3 play compensatory roles under certain conditions.^25–30 Mice with inactivation of PHD2 are embryonic lethal whereas PHD1 or PHD3 deficient mice are viable and normal.^27,28 siRNA-mediated knockdown of PHD2, but not PHD1 or PHD3, leads to HIF stabilization in cell culture.²⁵ Thus, we set to determine the fundamental role of PHD2 in regulating pulmonary vascular remodeling and the pathogenesis of severe PAH. We observed that PHD2 expression was diminished in endothelial cells (ECs) of obliterative pulmonary vessels of IPAH patients. Employing the mice with Tie2Cre-mediated disruption of Egln1 (encoding PHD2) in ECs and hematopoietic cells (HCs) (Egln1^Tie2), we observed spontaneous severe PAH with extensive pulmonary vascular remodeling including vascular occlusion and plexiform-like lesions, and severe right ventricular (RV) hypertrophy and right-sided heart failure. This phenotype is strikingly different from PH in mice induced by hypoxia alone or hypoxia and Sugen 5416 treatment.^{17, 31–33} To the best of our knowledge, this is the first mouse model with irreversible obliterative vascular remodeling and pathophysiology as seen in patients with severe PAH including IPAH.^{9,10, 32,33} We identified HIF-2α(not HIF-1α) activation as the critical mediator of severe PAH downstream of PHD2 deficiency. We also showed that HIF-2α-mediated expression of CXCL12 (also known as stromal cell-derived factor 1α, SDF-1α) in PHD2-deficient ECs promoted pulmonary arterial SMC proliferation and contributed to the pathogenesis of severe PAH in Egln1^Tie2 mice.

Methods

Mice

To generate Egln1^Tie2 mice, Egln1 floxed mice²⁸ were bred with Tie2 promoter/enhancer- driven Cre transgenic mice.³⁴ Egln1^Tie2/Hif1a ^Tie2 (EH1) and Egln1^Tie2/Hif2a^Tie2 (EH2) mice were generated by breeding Hif1a or Hif2a floxed mice with Egln1^Tie2 mice. Egln1^Tie2/Cxcl12^Tie2 (ECx) mice were generated by breeding Cxcl12 floxed mice with Egln1^Tie2 mice. All breeder mice were purchased from the Jackson Laboratory and littermate WT mice were used as controls. Both male and female mice were used for the experiments. The experiments were conducted according to NIH guidelines on the use of laboratory animals. The animal care and study protocol was approved by the Institutional Animal Care and Use Committee of the University of Illinois at Chicago.

Human subjects

Human lung tissues were obtained from patients undergoing lung transplantation for IPAH and from unused donor lungs. Informed consent from patients and local ethical approval from the ethics committees of the Hammersmith Hospitals (ref. no. 2001/6003) and Royal Brompton & Harefield Hospitals (ref. no. 01–210) were obtained prior to tissue collection. ³⁵

Statistical analysis

Statistical significance was determined by one-way or two-way ANOVA with a Games-Howell post hoc analysis that calculates P values corrected for multiple comparisons. Two- group comparisons were analyzed by the unpaired two-tailed Student’s t test for equal variance or Welch t test for unequal variance. Statistical analysis of the mortality study was performed with the Log–rank (Mantel-Cox) test. P < 0.05 denoted the presence of a statistically significant difference. All bar graphs represent mean ± SD. An expanded Materials and Methods section containing detailed description of echocardiography, RV hemodynamic measurement, bone marrow transplantation, histology and immunostaining, primary culture of human lung microvascular ECs and pulmonary arterial SMCs, siRNA-mediated knockdown, molecular analysis, RNA sequencing and bioinformatics analysis is provided in the online-only Data Supplement.

Results

Loss of PHD2 in endothelial cells and hematopoietic cells induces severe PAH and right heart hypertrophy and failure

To gain insight into the role of PHD2 in the pathogenesis of PAH, we inactivated PHD2 in the mouse endothelium. Mice carrying an Egln1 gene in which exons 2 and 3 were flanked by 2 loxP sites were bred with Tie2 promoter/enhancer-driven Cre (^Tie2Cre) transgenic mice to generate mice with Tie2Cre-mediated disruption of Egln1 (Egln1^Tie2) (Figure 1A). Western blotting analysis showed that PHD2 expression was reduced greater than 90% in lung ECs of Egln1^Tie2 mice (Figure 1B). Quantitative real-time RT-PCR (QRT-PCR) also revealed selective deletion of PHD2 in lung ECs but not fibroblasts (Figure 1C). Immunostaining demonstrated diminished PHD2 expression in pulmonary vascular ECs but not in bronchial epithelial cells in Egln1^Tie2 mice (Figure 1D). Together, these data demonstrate EC-specific deletion of PHD2 in Egln1^Tie2 mouse lungs. Immunostaining also showed loss of PHD2 expression in aortic ECs of Egln1^Tie2 mice (Supplemental Figure 1). Egln1^Tie2 mice were born normally in a Mendelian ratio. To determine whether these mutant mice develop PAH, we measured right-ventricular systolic pressure (RVSP), which reflects PA systolic pressure. Egln1^Tie2 mice had dramatically elevated RVSP that ranged from 60–90 mmHg in mice aged 3.5 months (Figure 1, E and F).

Figure 1.

Spontaneous severe PAH in Egln1^Tie2 mice. (A–D) Tie2Cre-mediated disruption of Egln1 in lung ECs. A diagram showing the strategy for generation of Egln1^Tie2 mice (A). Representative Western blotting demonstrating diminished PHD2 protein expression in EC lysate isolated from Egln1^Tie2 lungs compared to WT (B). The experiment was repeated twice with similar data. QRT-PCR analysis demonstrating PHD2 deletion in Egln1^Tie2 lung ECs but not in fibroblasts (Fib) (C). Data are expressed mean ±SD (n=3) (C). Representative micrographs of immunostaining showing EC-specific disruption of PHD2 in Egln1^Tie2 mouse lungs. Lung tissues were co-stained with anti-PHD2 and anti-CD31 (marker for ECs). Nuclei were counterstained with DAPI. PHD2 expression was diminished in pulmonary vascular ECs but not bronchial epithelial cells in Egln1^Tie2 lungs. Please note the marked vascular remodeling in Egln1^Tie2 lungs. Br, bronchiole; V, vessel. Scale bar, 50 µm (D). CKO = Egln1^Tie2. (E) Representative RVSP tracings (each=1 s). (F) Dramatic increase of RVSP in Egln1^Tie2 mice. Bars represent the mean. (G) Marked RV hypertrophy in Egln1^Tie2 mice. **, P < 0.01; ***, P < 0.001 (Student’s t test: C; Two-way ANOVA followed by Games-Howell post hoc analysis: F and G).

However, systemic blood pressure was normal (Supplemental Figure 1). Egln1^Tie2 mice also developed pronounced RV hypertrophy evident by the drastic increase of the weight ratio of right/left ventricle plus septum (RV/LV+S) which averaged 0.9, a value not seen in other mouse PH models (Figure 1G).

Given that PAH induces RV hypertrophy and heart failure, we performed echocardiography to evaluate cardiac size and function in vivo. The RV chamber was markedly enlarged in Egln1^Tie2 mice whereas barely evident in WT mice (Figure 2, A and B). Egln1^Tie2 mice exhibited marked RV hypertrophy as evidenced by a 3-fold increase in RV wall thickness (Figure 2C), and decreased RV contractility (Figure 2D). However, the LV cardiac size and function of Egln1^Tie2 mice were similar to those of WT mice (Supplemental Figure 2), suggesting that PAH in Egln1^Tie2 mice is not secondary to LV dysfunction. Egln1^Tie2 mice also exhibited a decreased ratio of PA acceleration time/ejection time (PA AT/ET) (Figure 2E), indicating PA diastolic dysfunction. To determine whether Egln1^Tie2 mice develop right-sided heart failure, we assessed expression of molecular markers of heart failure.^{36, 37} QRT-PCR analysis revealed a marked increase of expression of atrial natriuretic factor and skeletal α-actin (Figure 2F). Reactivation of this embryonic gene program is a common feature of heart failure. As seen in IPAH patients, Egln1^Tie2 mice also exhibited progressive mortality; 80% of Egln1^Tie2 mice died by the age of 6 months (Figure 2G). These findings demonstrate that Egln1^Tie2 mice develop spontaneous severe PAH which results in RV failure and premature death as seen in the clinical setting.

Figure 2.

Severe RV hypertrophy and failure and progressive mortality of Egln1^Tie2 mice. (A) Representative echocardiography showing an enlarged RV chamber and thickened RV wall (hypertrophy) in Egln1^Tie2 mice (3.5 mo old). (B–E) Echocardiography demonstrating enlarged RV chamber at end systole (ES) and end diastole (ED) (B); increased RV wall thickness diastole (RVWTD) (C); decreased RV fraction area change (RVFAC), indicative of decreased RV contractility (D); and a decreased PA AT/ET ratio (E) in Egln1^Tie2 mice (3.5 months old). Data are expressed as mean ± SD (n=5 WT and 6 Egln1^Tie2). (F) QRT-PCR analysis demonstrating reactivation of an embryonic gene program in the right ventricles of Egln1^Tie2 mice aged 3.5 mo, indicative of heart failure. ANF=Atrial natriuretic factor; sk-actin=skeletal α-actin. (G). Progressive mortality of Egln1^Tie2 mice. *, P < 0.05; **, P < 0.01; ***, P < 0.001. Student’s t test (B-E), Welch t test (F) and log-rank (Mantel-Cox) test (G) were used for statistical analysis.

Because Tie2Cre also induces gene deletion in hematopoietic cells (HCs) besides ECs,^38,39 we next determined the contribution of PHD2-deficient HCs. We transplanted Egln1^Tie2 bone marrow cells to lethally irradiated WT mice (WT chimeric mice) and WT bone marrow cells to lethally irradiated Egln1^Tie2 mice (Egln1^Tie2 chimeric mice). As shown in Figure 3, WT chimeric mice reconstituted with Egln1^Tie2 bone marrow cells did not show evidence of PAH with normal RVSP and RV/(LV+S) ratio. However, Egln1^Tie2 chimeric mice reconstituted with WT bone marrow cells had decreased RVSP and RV/(LV+S) ratio compared to Egln1^Tie2 mice. Pulmonary vascular remodeling including adventitial and intimal but not medial thickness in Egln1^Tie2 chimeric mice was also attenuated (Supplemental Figure 3). These data suggest that endothelial PHD2 deficiency is essential for induction of PAH whereas PHD2 deletion in HCs promotes the severity of PAH.

Figure 3.

Attenuated PAH in Egln1^Tie2 chimeric mice reconstituted with WT bone marrow cells. (A) A cartoon showing bone marrow (WT or CKO BM) transplantation to lethally irradiated WT mice. (B) QPCR analysis of Sry sequence (Y chromosome-specific gene) demonstrating greater than 95% efficiency of bone marrow reconstitution of female recipient WT mice with male Egln1^Tie2 bone marrow. Bone marrow DNA from male WT mice was used as positive control. (C, D) Reconstitution of lethally irradiated WT mice with bone marrow cells from Egln1^Tie2 mice (CKO BM) had no effects on RVSP (C) and RV/LV+S ratio (D). As controls, WT bone marrow cells were also transplanted to WT recipient mice. (E-G) Egln1^Tie2 chimeric mice reconstituted with WT bone marrow cells exhibited marked decreases of RVSP and RV hypertrophy. ***, P < 0.001 (Student’s t test). CKO=Egln1^Tie2.

Obliterative pulmonary vascular remodeling in Egln1^Tie2 mice

To further compare the similarities between the Egln1^Tie2 mouse phenotype and clinical PAH, we examined pulmonary pathology. Lung sections from Egln1^Tie2 mice (3.5 mo old) showed various forms of vascular remodeling, including pulmonary arterial intima, medial and adventitial thickness, occlusive intima and plexiform-like lesions (Figure 4, A and B and Supplemental Figure 4). Occlusion was observed in both large and small pulmonary arteries (Figure 4A and Supplemental Figure 4). Anti-CD31 (marker for ECs) immunohistochemistry revealed strong CD31 staining in ECs of pulmonary vessels without complete occlusion (Figure 4B and Supplemental Figure 4, I and J) but diminished in completely occluded vessels (Supplemental Figure 4, I and J). Intriguingly, multi-CD31⁺ channels were evident in the adventitial of some severely remodeled pulmonary vessels (Figure 4B). Histological quantification revealed more severe pulmonary vascular remodeling in Egln1^Tie2 mice at age of 3.5 mo compared to age of 1.5 mo. In Egln1^Tie2 mice aged 3.5 mo, 16% of the vessels were occlusive and 8% of large vessels (>100 µm diameter) exhibited plexiform-like lesions including lesions with multi-CD31⁺ channels (Figure 4C). Prominent immunostaining of smooth muscle α-actin was evident in theses lesions (Figure 4D). Immunostatining with anti-FSP1 (marker for fibroblasts) or anti-CD11b (marker for monocytes) demonstrated the adventitial of these lesions was fibroblast-enriched whereas fewer CD11b-positive monocytes (Figures 4, E and F).

Figure 4.

Occlusive pulmonary vascular remodeling in Egln1^Tie2 mice. (A) Representative micrographs of Russel-Movat pentachrome staining demonstrating thickening of the intima, medial, and adventitial, occlusion of the large and small vessels (black arrowheads), and plexiform-like lesions (red arrowheads) in 3.5 mo old Egln1^Tie2 mice (n=3 WT and 6 Egln1^Tie2 mice). Br, bronchus; V, vessel. Scale bar: 50 µm. (B) Anti-CD31 immunohistochemistry showing multiple-channel lesions positive for the endothelial marker CD31 (arrows). Scale bar: 50 µm. (C) Quantification of pulmonary vascular remodeling in Egln1^Tie2 mice. Grade 1 (G1), medial hypertrophy; G2, medial hypertrophy and intimal thickening (partial occlusion); G3, occlusive lesions (>75% occlusion); G4, plexiform-like lesions. Much severe pulmonary vascular remodeling was identified in Egln1^Tie2 mice at age of 3.5 mo compared to 1.5 mo. Total= all vessels quantified (n=280 vessels/group, n=5 mice/group). In 3.5 mo old Egln1^Tie2 mice, plexiform-like lesions (G4) were mainly seen in large vessels (>100 µm diameter) whereas occlusive lesions (G3) were prominent in small vessels. (D-F) Representative micrographs showing anti-smooth muscle α-actin (α-SMA), anti-FSP1 or CD11b staining in pulmonary vascular lesions of Egln1^Tie2 mice. . Br, bronchus; V, vessel. Scale bar, 50µm. (G) Representative micrographs showing proliferating ECs, SMCs and adventitial cells in pulmonary vascular lesions of Egln1^Tie2 mice (3.5 mo old). Lung sections were immunostained with anti-Ki67 for cell proliferation and anti-CD31 for ECs. Arrowheads point to proliferating ECs; arrows indicate proliferating SMCs; open arrows denote proliferating adventitial cells. Scale bar, 50 µm. (H) Western blotting demonstrating increased expression of PCNA, a marker for cell proliferation in Egln1^Tie2 lungs.

Increased cell proliferation in pulmonary vascular lesions in Egln1^Tie2 mice

We next determined whether vascular occlusion and formation of plexiform-like lesions in Egln1^Tie2 lungs was associated with abnormal vascular cell proliferation. Mouse lung sections were immunostained with anti-Ki67, a marker for cell proliferation and anti-CD31 for ECs. As shown in Figure 4G, Ki67 staining was observed in pulmonary vascular ECs, SMCs as well as adventitial cells in Egln1^Tie2 lungs but not evident in control WT lung sections. Furthermore, Western blotting demonstrated a marked increase of expression of proliferating cell nuclear antigen (PCNA), a marker for cell proliferation in Egln1^Tie2 lung tissues (Figure 4H).

Diminished expression of PHD2 in ECs of occlusive pulmonary vessels in IPAH patients

To further determine the clinical relevance of our observations in Egln1^Tie2 mice, we assessed PHD2 expression in lung sections from IPAH patients and normal donors.³⁵ PHD2 expression was diminished in the lumen (ECs) of occlusive pulmonary vessels of IPAH patients (Figure 5). However, non-occlusive pulmonary vessels from the same patients exhibited relatively normal PHD2 expression as seen in the control samples (Figure 5). These data suggest an important role for PHD2 deficiency in pulmonary vascular ECs in mediating obliterative vascular remodeling in patients.

Figure 5.

Diminished PHD2 expression in occlusive pulmonary vessels of IPAH patients. (A) Immunostaining demonstrating diminished PHD2 expression (red) in the lumen of occlusive vessels (arrowheads) of IPAH lungs. Lung sections exhibited strong autofluorescence (AutoF) which helped to show the morphology. Arrows point to non-occlusive vessels; asterisk indicates blood cells. Scale bar, 50 µm. (B) Quantification of PHD2 expression. Immunofluorescent intensity (Fluo) was graded from 1 to 10 with 10 the highest. PHD2 expression was diminished in occlusive vessels of IPAH patients. Data are expressed as mean ± SD. **, P < 0.01 (Welch t test). A.U., arbitrary units.

HIF-2α activation secondary to PHD2 deficiency mediates PAH in Egln1^Tie2 mice

To determine the molecular basis of severe PAH in Egln1^Tie2 mice, we first analyzed HIF- 1α and HIF-2α protein expression. Both proteins were markedly increased in Egln1^Tie2 lungs, a result consistent with PHD2 deficiency-induced stabilization (Figure 6A). We then generated Egln1^Tie2/Hif1a^Tie2 (EH1) and Egln1^Tie2/Hif2a^Tie2 (EH2) double knockout mice by breeding Hif1a or Hif2a floxed mice with Egln1^Tie2 mice (Figure 6B). The RVSP and the RV/LV+S ratio in EH2 mice were completely normal (Figure 6, C and D, and Supplemental Figure 5), whereas EH1 mice exhibited similar changes to those in Egln1^Tie2 mice (Figure 6, C and D).

Figure 6.

Role of HIF-2α activation in Egln1^Tie2 mice in mediating severe PAH. (A) Western blot showing stabilized HIF-1α and HIF-2α expression in Egln1^Tie2 mouse lungs. The experiment was repeated twice with similar data. (B-D) Genetic deletion of HIF2α but not HIF1α in Egln1^Tie2 mice completely normalized RVSP (C) and inhibited RV hypertrophy evident by normalized RV/LV+S ratio (2 months old). (D) Bars represent the mean. (E, F) Echocardiography demonstrating normalization of the RVWTD (E) and the PA AT/ET ratio (F) in EH2 mice at age of 3.5 months. Data are expressed as mean ± SD (n=6 WT, 8 Egln1^Tie2 and 5 EH2). (G) Russel-Movat pentachrome staining showed normal pulmonary vascular structure in EH2 mice at 3.5 month of age in contrast to Egln1^Tie2 mice. Br, bronchiole; V, vessel. Scale bar: 50 µm. (H) Representative micrographs showing increased muscularization of distal pulmonary vessels (<50µm in diameter) in Egln1^Tie2 mice and normalization in EH2 mice aged 3.5 months. Lung sections were immunostained with anti-smooth muscle α-actin (SMA) (red). Arrows point to muscularized distal vessels. Scale bar, 50 µm. (I) Quantification of muscularization. SMA– positive vessels were counted in 40 fields (×200) from each lung section. Data are expressed as mean ± SD (n=5 mice/group). *, P < 0.05; ***, P < 0.001. ANOVA followed by Games-Howell post hoc analysis was used for statistical analysis.

Echocardiography also revealed normal RV wall thickness indicating no hypertrophy, as well as normal PA diastolic function evident by normal value of AT/ET ratio in EH2 mice (Figure 6, E and F). EH2 mice at 3.5 mo of age also exhibited normal pulmonary vascular pathology without evidence of vascular wall thickening, occlusion, or muscularization of the distal pulmonary arteries (Figure 6, G–I). These data demonstrate that activation of HIF-2α (not HIF-1α) secondary to PHD2 deficiency is responsible for the obliterative vascular remodeling and severe PAH in Egln1^Tie2 mice.

Altered expression of PH-causing genes in Egln1^Tie2 lungs and normalization in EH2 lungs

To delineate the molecular signaling events that contribute to the severe PAH phenotype of Egln1^Tie2 mice downstream of HIF-2α, we performed whole transcriptome RNA sequencing (RNA-seq) analysis (Figure 7, A–E). RNA-seq data analysis showed up-regulation of HIF-α target genes in Egln1^Tie2 lung, consistent with activation of HIF-α. Expression of some of them such as Arg1, Egln3, Serpine1, pak6, Il6, Lox, and Nov was completely normalized in EH2 lungs, suggesting these are HIF-2α target genes in Egln1^Tie2 lungs (Figure 7B). We also analyzed the expression profiles of genes involved in the pathogenesis of PH. Genes shown upregulated in various animal models of PH was markedly induced in Egln1^Tie2 lungs. Expression of these genes was normalized in EH2 lungs (Figure 7C). Expression of genes known downregulated in published PH models was decreased in Egln1^Tie2 lungs and restored to levels either similar to or greater than WT lungs in EH2 mice (Figure 7, D and E). QRT-PCR analysis confirmed marked changes of expression of these genes in Egln1^Tie2 lungs, including increased expression of Slc39a12, Eln, Sphk1, Csf2, Cxcl12, Il6, and Edn1 and decreased expression of Ccr7, Apln, Aplnr, Cav1, Prkg1, Ccr2, and Bmpr2. Expression of these genes was normalized in EH2 lungs (Figure 7, F and G). These data suggest the complex pathology and severe PAH in Egln1^Tie2 mice are attributable to dysregulation of multiple signaling pathways secondary to PHD2 deficiency-activated HIF-2α.

Figure 7.

Expression profiling of genes associated with various PH animal models. (A) Representative heat map of RNA-seq analysis in WT, Egln1^Tie2 (CKO) and EH2 mouse lungs (n=3 mice/group). (B-D) RNA-seq analysis of HIF-α target genes (B), PH-causing genes (C, D) in mouse lungs. (E) Representative diagram showing increased Cxcl12 expression in Egln1^Tie2 lungs whereas inhibited in EH2 lungs identified by RNA-seq analysis. (F) QRT-PCR analysis of expression of genes known upregulated in published PH animal models. Data are expressed as mean ± SD (n=4 WT, 6 CKO and 6 EH2). (G) QRT-PCR analysis of expression of genes known downregulated in PH animal models. *, P < 0.05; **, P < 0.01; ***, P < 0.001. ANOVA followed by Games-Howell post hoc analysis was used for statistical analysis.

PHD2-deficient ECs induces pulmonary arterial SMC proliferation partially through HIF-2α-dependent release of CXCL12

Given that pulmonary arterial SMC proliferation plays an important role in pulmonary vascular remodeling, we next determined the role of PHD2 deficiency in ECs in directing pulmonary arterial SMC proliferation. Employing a co-culture Transwell system, human lung microvascular ECs deficient in PHD2 induced by PHD2 siRNA-mediated knockdown (Figure 8A) were plated in the top chamber while human pulmonary arterial SMCs were cultured in the bottom chamber. As shown in Figure 8, B and C, PHD2-deficient ECs induced a marked increase of SMC proliferation, suggesting PHD2-deficient ECs play an important role in pulmonary vascular remodeling by directly releasing factor(s) to activate SMC proliferation.

Figure 8.

HIF-2α-mediated induction of CXCL12 in lung ECs induced SMC proliferation and contributed to severe PAH in Egln1^Tie2 mice. (A) QRT-PCR analysis demonstrating HIF- 2α-dependent induction of CXCL12 in PHD2-deficient ECs. Small interfering RNAs specific for either PHD2, HIF1A, or HIF2A were transfected to human lung microvascular ECs. Scramble RNA against PHD2 was also transfected as a control. Data are expressed as mean ± SD (n=3). (B) Representative micrographs showing PHD2-deficient ECs induced SMC proliferation employing the Transwell co-culture system. Proliferating SMCs were immunostained with anti-BrdU antibody (green). Scale bar, 50 µm. (C) Quantification showing PHD2-deficient ECs induced a marked increase of SMC proliferation which was partially inhibited by knockdown of CXCL12. n=3. (D) QRT-PCR analysis demonstrating siRNA- mediated knockdown of CXCL12 in human lung ECs. (E) Cxcl12 expression was markedly induced in Egln1^Tie2 (CKO) lungs in a HIF-2α but not HIF-1α-dependent manner. Lung tissues were collected from 1.5 mo old mice. (F) Generation of a mouse model with genetic deletions of both Egln1 and Cxcl12 (ECx). (G) ECx mice exhibited markedly decreased RVSP compared to CKO mice (3.5 mo old). (H) RV hypertrophy was also partially inhibited in ECx mice. **, P < 0.01; ***, P < 0.001. ANOVA followed by Games-Howell post hoc analysis was used for statistical analysis (A, C, E). Student’s t test was used for statistical analysis (D, G, H). (I) A diagram showing the model that PHD2 deficiency in ECs and HCs induces obliterative pulmonary vascular remodeling and severe PAH. Endothelial PHD2 deficiency is essential for the development of PAH while PHD2 deficiency in HCs promotes the severity of the disease. Through HIF-2α activation, PHD2 deficiency induces dysregulation of multiple signaling pathways in mouse lungs which may act together to induce severe pulmonary vascular remodeling including vascular occlusion and formation of plexiform-like lesions as well as vasoconstriction and thereby severe PAH and RV failure.

We next explored the role of CXCL12 released by PHD2-deficient ECs in promoting SMC proliferation. As shown in Figure 8A, siRNA-mediated knockdown of PHD2 in human lung microvascular ECs induced a five-fold increase of expression of CXCL12. HIF-2α siRNA- mediated knockdown of HIF-2α inhibited PHD2 deficiency-induced CXCL12 expression whereas HIF-1α knockdown had no effect, demonstrating CXCL12 is a HIF-2α but not HIF-1α target gene in PHD2-deficient lung ECs. To determine whether induced CXCL12 expression contributes to SMC proliferation, we employed siRNA to knockdown CXCL12 in PHD2- deficient ECs (Figure 8D). Co-culture experiment demonstrated knockdown of CXCL12 in PHD2-deficient ECs resulted in a marked decrease of SMC proliferation (Figure 8, B and C), suggesting CXCL12 is one of the factors released from PHD2-defieicnt ECs promoting SMC proliferation.

PHD2/HIF-2α-dependent activation of CXCL12 signaling contributes to the pathogenesis of severe PAH in Egln1^Tie2 mice

In Egln1^Tie2 lungs, we also observed a marked increase of CXCL12 expression compared to WT lungs which was inhibited in EH2 lungs but not in EH1 lungs (Figure 8E), consistent with our observation in PHD2-deficient ECs described above. Thus, we generated a double knockout mouse model with genetic deletions of both Egln1 and Cxcl12 by breeding Cxcl12 floxed mice into the genetic background of Egln1^Tie2 mice (Figure 8F). RVSP measurement revealed a marked decrease of PAH in Egln1^Tie2/Cxcl12^Tie2 mice compared to Egln1^Tie2 mice (Figure 8G). RV hypertrophy was also markedly attenuated in these double knockout mice (Figure 8H). Together, these data demonstrate that HIF-2α-dependent activation of CXCL12 signaling contributes to the pathogenesis of severe PAH in Egln1^Tie2 mice (Figure 8I).

Discussion

The present study has demonstrated for the first time that genetic deletion of Egln1 in ECs and HCs induces spontaneous progressive PAH with severe vascular remodeling including occlusion and complex plexiform-like lesions that resemble the pathology of clinical PAH. As seen in IPAH patients, Egln1^Tie2 mice exhibit severe RV hypertrophy and failure and thereby progressive mortality. We also observed dysregulated expression of multiple PH-causing genes in Egln1^Tie2 lungs, and demonstrated that CXCL12 induced in PHD2-deficient ECs promoted SMC proliferation. Inhibition of CXCL12 by Tie2Cre-mediated genetic deletion attenuated PAH in Egln1^Tie2 mice. These complex pathological and molecular changes in Egln1^Tie2 lungs are ascribed to HIF-2α activation. The observation that PHD2 expression is diminished in ECs of obliterative pulmonary vessels of IPAH patients further demonstrates the clinical relevance of this model. These findings may lead to novel therapeutic strategies for the treatment of severe PAH including IPAH in patients.

Vascular occlusion and complex plexiform lesions are hallmarks of the pathology of severe PAH in patients. Although PAH in rats treated with vascular endothelial growth factor receptor inhibitor Sugen 5416 and chronic hypoxia shows severe pulmonary vascular remodeling with stable vascular occlusion and plexiform-like lesions,^40,41 obliterative vascular lesions in lungs of hypoxia/Sugen 5416-treated mice are not stable and resolved after 10 weeks of normoxic exposure.³³ A recent study also shows occlusive lung arterial lesions in a transgenic mouse model with inducible EC apoptosis.⁴² However, the majority of these transgenic mice exhibit normal lung phenotype and only a small proportion (21%) of them develop mild PAH with scarce occlusive lesions. Other genetically modified mouse models don’t exhibit such complex vascular lesions.^9,10 RV hypertrophy is in general mild to median (RV/LV+S ratio = 0.3–0.5) in these alternative mouse models. Thus, there is no PAH mouse model that recapitulates the complex pathology of clinical PAH in the literature.^9,10 Our study provides unequivocal evidence that Tie2Cre-mediated disruption of Egln1 induces spontaneous severe PAH with characteristics of irreversible vascular occlusion and plexiform-like lesions. These mice exhibit marked increase of RVSP ranging from 60–90 mmHg and unprecedented RV hypertrophy with RV/LV+S ratio averaged at 0.9. To the best of our knowledge, this is the first mouse model showing such great increases of RVSP and severe RV hypertrophy. Consistent with the progressive mortality seen in IPAH patients, these mice also exhibit progressive mortality and 80% of them die by age of 6 months. It is likely that these mice die of right heart failure as they exhibit severe RV hypertrophy, decreased RV contractility, and reactivation of an embryonic gene program in the RV, a common feature of human heart failure.³⁶ Intriguingly, Egln1^Tie2 mice also exhibit altered expression of many genes involved in the pathogenesis of PH. Among the 21 genes validated by QRT-PCR analysis, 18 genes are dysregulated in Egln1^Tie2 lungs. Together, these data demonstrate the first mouse model recapitulating the pathogenesis of clinical PAH. Additionally, PHD2 expression is diminished in ECs of obliterative pulmonary vessels in IPAH patients, further demonstrating the clinical relevance of this model.

It is well known that hypoxia and HIF signaling play an important role in the pathogenesis of PH. Unexpectedly, we show spontaneous severe PAH with complex pulmonary vascular remodeling in Egln1^Tie2 mice. The severity and complexity of the pulmonary phenotype have not been reported in mouse models induced by either hypoxia or hypoxia/Sugen 5416 or genetic modifications of other HIF signaling molecules. Egln1^Tie2 mice have much more severe PAH phenotype than either Vhl^R200W mutation mice or Hif2a ^G536W knockin mice (average RVSP: Egln1^Tie2 mice at age of 3.5 months, 70 mmHg; Hif2a ^G536W knockin mice at 4–6 months age, 42mmHg). There are no occlusive lesions but only slight increase of small pulmonary arterial thickness in Vhl^R200W mutation mice or Hif2a ^G536W knockin mice.^{22, 24} Intriguingly, the severe PAH phenotype in Egln1^Tie2 mice is ascribed to activation of HIF-2α but not HIF-1α. Both pathological and molecular changes in Egln1^Tie2 mice are normalized by genetic deletion of Hif2a. Consistent with our observation, previous studies have shown that heterozygous Hif2α deletion partially rescues Vhl^R200W mutation-induced PAH.²⁴ Together, these studies point out an obligatory role of HIF-2α activation in the pathogenesis of PAH. Although HIF-1α and HIF-2α bind to an identical core sequence in hypoxia response element, they can activate distinct sets of genes.⁴³ Data from our RNA sequencing analysis show expression of some of the HIF-α target genes are dysregulated in Egln1^Tie2 mouse lungs but normalized by Hif2a deletion, suggesting these are HIF-2α target genes whereas a couple of them such as Bnip3 and Pgk1 are more likely HIF-1α target genes. In human lung ECs, our data demonstrate CXCL12 as a target gene of HIF- 2α. Additionally, HIF-2α is predominantly expressed in ECs whereas HIF-1α is more ubiquitously expressed almost in all cell types.⁴⁴ Thus, HIF-2α may function differently than HIF-1α under various (patho) physiological conditions.

Pulmonary endothelial dysfunction has been identified as a critical mediator of pulmonary vascular remodeling. Our co-culture study provides clear evidence that PHD2-deficient ECs directly induce SMC proliferation, indicating the importance of endothelial crosstalk with SMCs in the mechanisms of pulmonary vascular remodeling. We have identified CXCL12 as one of the factors released from PHD2-deficient ECs to induce SMC proliferation. The partial inhibition of SMC proliferation by CXCL12 knockdown in PHD2-deficient ECs suggest other angiogenic factor(s) also contribute to EC-induced SMC proliferation. Consistent with our observation in vitro, genetic deletion of Cxcl12 results in marked decrease of PAH in Egln1^Tie2 mice. Our data from RNA sequencing as well as QRT-PCR analysis demonstrate dysregulation of many PH- causing genes including Edn1 (encoding Endothelin-1) and Prkg1 (encoding protein kinase G) in Egln1^Tie2 lungs. Thus, it is unlikely to completely rescue the hypertensive pulmonary phenotype in Egln1^Tie2 mice by targeting one specific pathway downstream of PHD2/HIF-2α signaling.

These data also suggest the importance of combination therapy in treatment of PAH in patients.⁴⁵ Another surprising finding is that PHD2 deficiency induced severe PAH but had no effects on systemic pressure. These differential effects of PHD2 deficiency are likely due to the marked difference of expression of PHD2 in the pulmonary vasculature compared to the systemic vasculature. PHD2 is highly expressed in pulmonary vascular ECs but much weakly in systemic vascular ECs. Accordingly, PHD2 deficiency-induced expression of Cxcl12 and Endothelin-1 in pulmonary vascular ECs and Egln1^Tie2 mouse lungs was not seen in Egln1^Tie2 mouse aorta.

Besides the role Cxcl12 in the development of severe PAH in Egln1^Tie2 mice, Endothelin-1 is a potent vasoconstrictor and clinical studies have demonstrated its role in the pathogenesis of PAH. Together, these data provide clear evidence about the differential effects of PHD2 deficiency on the pulmonary vascular system and the systemic vasculature.

Our studies also demonstrate an important role of PHD2 deficiency in HCs in mediating the severe PAH phenotypes in Egln1^Tie2 mice. On one hand, WT chimeric mice reconstituted with Egln1^Tie2 bone marrow cells fail to develop PAH, indicating PHD2 deficiency in bone marrow cells is not a prerequisite condition for the development of PAH. On the other hand, reconstitution of WT bone marrow cells in lethally irradiated Egln1^Tie2 mice attenuates PAH, indicating Tie2Cre-meidated disruption of PHD2 in HCs contributes to the severity of PAH. These transplanted chimera experiments demonstrate the synergistic contribution of both endothelial and hematopoietic PHD2 deficiency to the severe PAH phenotype and complex pathology seen in Egln1^Tie2 mice. Prior studies also demonstrate an important role of bone marrow abnormality associated with altered HIF signaling in the pathogenesis of clinical PAH.^46,47 In addition, accumulating evidence shows that macrophages, T cells, and progenitor cells are recruited in the lesions contributing to vascular remodeling.^47–49 It has also been shown that HIF- stimulated extracellular adenosine production and signaling affect the pathogenesis of PH through modulation of macrophage functional polarization⁵⁰. Given the observation of induced CXCL12 expression in PHD2-deficient ECs, PHD2 deficiency in ECs may also promote bone marrow cell recruitment to the lung and thereby induces complex vascular remodeling. Thus, future studies are warranted to determine which bone marrow cell population(s) contribute to the severe PAH phenotype in Egln1^Tie2 mice and whether PHD2-deficient EC-released CXCL12 and/or other chemokine(s) promote bone marrow cell recruitment and thereby contributing to the severity of PAH.

In summary, we have demonstrated for the first time that genetic deletion of Egln1 in ECs and HCs induces spontaneous progressive PAH with severe vascular remodeling including occlusion and complex plexiform-like lesions and pronounced RV hypertrophy and RV failure that resemble the pathology of clinical PAH. We also identify HIF-2α as a critical mediator of severe PAH seen in Egln1^Tie2 mice. The severe PAH phenotype characterized with complex obliterative vascular remodeling and dysregulation of various signaling pathways as seen in clinical PAH makes the Egln1^Tie2 mouse an unique and potentially useful mouse model for delineating the molecular mechanisms underlying the complex vascular remodeling of clinical PAH. Thus, targeting dysregulated PHD2-HIF-2α signaling may represent a novel effective therapeutic strategy to inhibit pulmonary vascular remodeling for the treatment of severe PAH, including IPAH and thereby promote survival.

Clinical Perspectives.

Pulmonary arterial hypertension (PAH) is a progressive fatal disease, which without treatment leads to right-heart failure and death often within 2–3 years of diagnosis. Owing to the poor understanding of the underlying mechanisms of obliterative vascular remodeling, current therapies result in only modest improvement in morbidity and mortality. One of the major hurdles is the lack of appropriate murine models recapitulating the pathophysiology of severe PAH in patients. In this study, we have demonstrated for the first time that genetic deletion of Egln1 in ECs and HCs induces spontaneous progressive PAH with severe pulmonary vascular remodeling including occlusion and complex plexiform-like lesions that resemble the pathology of clinical PAH. As seen in idiopathic PAH (IPAH) patients, Egln1^Tie2 mice exhibit severe right ventricular hypertrophy and failure and progressive mortality. We also observed dysregulated expression of multiple PH-causing genes in Egln1^Tie2 lungs. These complex pathological and molecular changes in Egln1^Tie2 lungs are ascribed to HIF-2α activation. The severe PAH phenotype characterized with obliterative vascular remodeling and dysregulation of various signaling pathways as seen in clinical PAH makes the Egln1^Tie2 mouse an unique and potentially useful mouse model for delineating the molecular mechanisms underlying the complex vascular remodeling of clinical PAH. The observation that PHD2 expression is diminished in ECs of obliterative pulmonary vessels of IPAH patients further demonstrates the clinical relevance of this model. Thus, targeting dysregulated PHD2/HIF-2α signaling may represent a novel effective therapeutic strategy to reverse pulmonary vascular remodeling for the treatment of severe PAH, including IPAH and promote survival.