SM22α cell-specific HIF stabilization mitigates hyperoxia-induced neonatal lung injury

Abstract

Though survival rates for preterm infants are improving, the incidence of chronic lung disease of infancy, or bronchopulmonary dysplasia (BPD), remains high. Histologically, BPD is characterized by larger and fewer alveoli. Hypoxia-inducible factors (HIFs) may be protective in the context of hyperoxia-induced lung injury, but the cell-specific effects of HIF expression in neonatal lung injury remain unknown. Thus, we sought to determine whether HIF stabilization in SM22α-expressing cells can limit hyperoxia-induced neonatal lung injury. We generated SM22α-specific HIF-1α-stabilized mice (SM22α-PHD1/2^−/− mice) by cross-breeding SM22α-promotor-driven Cre recombinase mice with prolyl hydroxylase PHD1^flox/flox and PHD2^flox/flox mice. Neonatal mice were randomized to 21% O₂ (normoxia) or 80% O₂ (hyperoxia) exposure for 14 days. For the hyperoxia recovery studies, neonatal mice were recovered from normoxia for an additional 10 wk. SM22α-specific HIF-1α stabilization mitigated hyperoxia-induced lung injury and preserved microvessel density compared with control mice for both neonates and adults. In SM22α-PHD1/2^−/− mice, pulmonary artery endothelial cells (PAECs) were more proliferative and pulmonary arteries expressed more collagen IV compared with control mice, even under hyperoxic conditions. Angiopoietin-2 (Ang2) mRNA expression in pulmonary artery smooth muscle cells (PASMC) was greater in SM22α-PHD1/2^−/− compared with control mice in both normoxia and hyperoxia. Pulmonary endothelial cells (PECs) cocultured with PASMC isolated from SM22α-PHD1/2^−/− mice formed more tubes and branches with greater tube length compared with PEC cocultured with PASMC isolated from SM22α-PHD1/2^+/+ mice. Addition of Ang2 recombinant protein further augmented tube formation for both PHD1/2^+/+ and PHD1/2^−/− PASMC. Cell-specific deletion of PHD1 and 2 selectively increases HIF-1α expression in SM22α-expressing cells and protects neonatal lung development despite prolonged hyperoxia exposure. HIF stabilization in SM22α-expressing cells preserved endothelial cell proliferation, microvascular density, increased angiopoietin-2 expression, and lung structure, suggesting a role for cell-specific HIF-1α stabilization to prevent neonatal lung injury.

INTRODUCTION

Initially reported by Northway et al. (1) in 1967, bronchopulmonary dysplasia (BPD) is classically considered to result from postnatal lung injury from high concentrations of inspired oxygen, positive pressure ventilation, infection, and inflammation (1, 2). With therapeutic advances in neonatal care, including prenatal steroid administration, surfactant replacement therapy, and low volume ventilator strategy, the incidence of BPD in more mature (>28 wk) infants decreased but increased overall, owing to the survival of very low birth weight (VLBW) infants. Interestingly, in the VLBW infants with BPD, lung histology is characterized by fewer and larger alveoli without inflammation, so called “new BPD” (3).

New BPD highlighted the importance of pulmonary angiogenesis as a key driver of alveolarization (3, 4). Proangiogenic factors are decreased, and pulmonary angiogenesis is impaired in infants dying from BPD. In experimental models, pharmacological or genetic inhibition of angiogenesis impairs both vascular growth and alveolarization, mirroring the abnormalities of human BPD (5). Conversely, exogenous vascular endothelial growth factor (VEGF) administration can enhance endothelial survival and function and mitigate the lung damage caused by neonatal hyperoxia (6), a commonly used preclinical model to study BPD (4).

Increasing evidence supports a role for hypoxia-inducible factors (HIFs) in promoting postnatal pulmonary angiogenesis and alveolarization. Downstream targets of HIF, including VEGF, are proangiogenic (7, 8). Distinct HIF isoforms, HIF-1α and HIF-2α, possess distinct expression patterns in the lung suggesting isoform- and cell-specific functions (9, 10). Genetic deletion of either isoform has significant implications with fatal cardiovascular malformations, compromised lung development, and neonatal respiratory distress (11, 12). In preterm sheep and primates, HIF-1α and HIF2α protein levels are markedly reduced in the lungs (13, 14). HIFs are constitutively expressed heterodimers comprised of an α- and a β-subunit (15). In normoxia, the α-subunit is rapidly hydroxylated prolyl hydroxylase enzymes and subsequently degraded (16). HIF stabilization using prolyl hydroxylase inhibitors can preserve lung structure, function, and growth despite injurious stimuli such as hyperoxia and mechanical ventilation (17). Recent data demonstrate that HIF stabilization can mitigate the adverse effects of intrauterine inflammation on placental structure and subsequent lung growth (18). Thus, although the importance of HIFs in lung development is clear, the isoform- and cell-specific role for these transcription factors remains unknown.

To gain further insight into the cell- and isoform-specific role of HIF in the neonatal lung, we created a mouse with constitutively stabilized HIF in SM22α-expressing cells by crossing SM22α-promotor-driven Cre recombinase mice with prolyl hydroxylase domain; PHD1^flox/flox and PHD2^flox/flox mice. As PHD2 is the major prolyl hydroxylase for HIF-α and PHD3 is most well expressed in cardiac myocytes, we opted for cell-specific deletion of PHD1 and 2 (19, 20). Animals with PHD1–3 deletion in SM22α-expressing cells did not survive to term. We determined that HIF-1 is the predominant isoform in SM22α-expressing cells and that constitutive HIF-1α stabilization protects the neonatal lung from hyperoxia-induced lung injury.

METHODS

Generation of SM22α-PHD1/2^−/− Mice

Transgenic mice with selective deletion of PHD1 and PHD2 (SM22α-PHD1/2^−/− mice) in SM22α-expressing cells in a C57BL/6 background were created by cross breeding SM22α-promoter-driven Cre recombinase mice with PHD1^flox/flox and PHD2^flox/flox mice. PHD1^flox/flox and PHD2^flox/flox mice were kindly provided by Dr. Amato Giaccia, Oxford University (21). The Institutional Animal Care and Use Committee at Stanford University approved all the procedures and protocols governing the care and use of laboratory animals.

To identify the Cre gene, the neonatal mice were genotyped for Cre using the following primers: forward 5′- CCATCTGCCACCAGCCAG-3′; reverse 5′- TCGCCATCTTCCAGCAGG-3′ generating a 300 bp PCR product (22). To identify the floxed PHD1 and PHD2 genes, the following primers were used: PHD1 forward 5′- TGGGCGCTGCATCACCTGTATCT-3′ reverse 5′- ACTGGTGAAGCCTGTAGCCTGTC-3′ and PHD2 forward 5′- CAAATGGAGATGGAAGATGC-3′; reverse 5′- TCAACTCGAGCTGGAAACC-3′ (The Jackson Laboratory protocol) generating 900 bp and 800 bp PCR products, respectively. Mice negative for Cre expression (SM22α-PHD1/2^+/+ mice) were used as controls.

Hyperoxia and Hyperoxia Recovery Studies

Within 24 h of birth, neonatal mice together with their dams were randomized to receive 21% O₂ (normoxia) or 80% O₂ (hyperoxia) exposure for 14 days. For the hyperoxia recovery studies, neonatal mice were treated as described and then allowed to recover in 21% O₂ (normoxia) for an additional 10 wk (hyperoxia recovery) (23). Continuous 80% O₂ exposure was achieved in a Plexiglas chamber by a flow-through system and the oxygen levels inside the Plexiglas chamber were continuously monitored with an oxygen sensor. Nursing dams were rotated between hyperoxia and normoxia every 24 h to prevent oxygen toxicity in the dams.

Lung Histology and Morphometry

Mice were euthanized at postnatal day 14 (P14) or at 12 wk of age by inhalation of carbon dioxide gas, and the trachea was cannulated with a 24-gauge outer-diameter polyethylene tube. The lungs were inflation-fixed through the trachea with 4% paraformaldehyde at 20 cmH₂O of gravity pressure for 5 min. The trachea was tied off, the lungs were removed, and then fixed in 4% paraformaldehyde overnight at 4°C. Fixed lung tissues were dehydrated in 70% ethanol, paraffin-embedded, and processed into 5-μm sections. Sections were stained by hematoxylin and eosin (H&E).

For radial alveolar count (RAC) assessment, 40 randomly chosen terminal respiratory bronchioles were examined at ×100 magnification by a blinded observer with a Leica DM5500 upright microscope and a Micropublisher 5MPixel color digital camera (Leica Microsystems, Buffalo Grove, IL). RAC was measured using Metamorph software (Leica Microsystems). In brief, the number of distal air sacs that were transected by a line drawn from a terminal respiratory bronchiole to the nearest pleural surface was counted. RAC results represent the average number of distal air sacs per genotype and condition.

For mean linear intercept (MLI) assessment, 10 randomly chosen images were examined at ×200 magnification by a blinded observer with a Leica microscope using Metamorph software (Leica Microsystems). In brief, the free distance in air space size (in µm) by a set of random line intercepts drawn across each field within the parenchyma was measured. MLI results represent the average air space size of alveoli per genotype and condition.

Primary Mouse PASMC and PEC isolation

Pulmonary artery smooth muscle cells (PASMCs) were isolated from control (SM22α-PHD1/2^+/+) and SM22α-PHD1/2^−/− adult mice, and pulmonary endothelial cells (PECs) were isolated from control P14 mice using a modified elastase/collagenase digestion protocol (24). For PASMC isolation, PA tissue was digested in dispersion medium containing 40 μmol/L CaCl₂, 0.5 mg/mL elastase (Worthington Biochemical), 0.5 mg/mL collagenase (Worthington Biochemical), 0.2 mg/mL soybean trypsin inhibitor (Worthington Biochemical), and 2 mg/mL albumin (Sigma Aldrich) for 30 min at 37°C. After filtration with 100-μm cell strainers, cells were incubated with Dynabeads (Invitrogen) coated with CD31 antibody (BD Biosciences) for 30 min, to adhere endothelial cells expressing CD31. Remaining smooth muscle cells (SMCs) were collected through centrifugation at 225 g for 5 min at 4°C and cultured in DMEM containing 10% FBS, 1% l-glutamine, and 1% antibiotic-antimycotic solution (Invitrogen/Gibco). To confirm isolation of PASMC, cells were stained for α-smooth muscle actin (α-SMA) (1:400, Sigma Aldrich A2547) by immunofluorescence. The percentage of cells positive for α-SMA expression was 100% for both genotypes. For PEC isolation, P14 lung tissue was treated as described earlier, with the following exceptions: isolated PEC was cultured in extracellular matrix (ECM) containing 10% FBS and 1% antibiotic-antimycotic solution (Invitrogen/Gibco). All experiments were performed with cells at passages 2–5.

Immunocytochemistry

PASMC isolated from control and SM22α-PHD1/2^−/− mice were fixed with 3% paraformaldehyde in PBS for 30 min, blocked and permeabilized in VSVG blocking solution (0.1% Triton X-100, 15 mg/mL glycine, 2.5% FBS in PBS) overnight at 4°C, and then incubated with HIF-1α antibody (1:400, GeneTex GTX127309, Irvine, CA) for 1 h in VSVG blocking solution. Cells were then incubated with Alexa Fluor 568 goat α-rabbit antibody (1:200, Thermo Fisher Scientific, Waltham, MA) for 1 h, followed by mounting in Vectashield Mounting Medium with 4′,6-diamino-2-phenylindole (DAPI) (Vector Laboratories, Burlingame, CA). Representative images are shown at ×200 magnification with results representing the average percentage of nuclear HIF-1α-expressing cells with a minimum of 100 cells counted per sample.

Western Immunoblotting

PASMC were incubated in normoxic or hyperoxic conditions (80% O₂, overnight) and then lysed with 0.5% NP-40 buffer [50 mM Tris·HCl, pH 7.5; 150 mM NaCl; 2.5 mM EDTA; 0.5% NP-40; 1 mM Na₃VO₄; 1 mM PMSF; 10 μg/mL aprotinin; 10 μg/mL leupeptin; and 5 μM HIF prolyl hydroxylase inhibitor (Calbiochem, San Diego, CA)]. Protein content was measured using the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific). Total proteins (15 μg/sample) were fractionated by SDS-PAGE on 4%–12% Tris-glycine precast gradient gels (Invitrogen) and then transferred to Immobilon-P membranes (Millipore, Billerica, MA). The membranes were incubated overnight at 4°C with respective primary antibodies to detect HIF-1α (Novus Biologicals NB100-105, Centennial, CO), PHD1 (Abcam ab108980), PHD2 (Santa Cruz Biotechnology sc-271835), and β-actin (Sigma Aldrich A5441), and then incubated for 1 h at room temperature with horseradish peroxidase-conjugated secondary antibodies. Antibody-bound proteins were detected by the ECL chemiluminescence methodology (GE Healthcare Life Sciences, Marlborough, MA). The intensities of protein bands were quantified by Quantity One Imaging Analysis Program (Bio-Rad), normalized for loading using β-actin as a control, and expressed as relative expression.

Pulmonary Vessel Density

Immunofluorescence staining for von Willebrand factor (VWF), an endothelial-specific marker, was performed for assessing vessel density; n = 5 or 6 per genotype and condition. The paraformaldehyde-fixed and paraffin-embedded lung sections from control and SM22α-PHD1/2^−/− mice were deparaffinized and rehydrated, incubated with Universal Antigen Retrieval Reagent (R&D Systems, Minneapolis, MN) for 30 min at 95°C, permeabilized with 0.25% Triton X-100-PBS solution for 30 min, incubated with 100 mM glycine solution (pH7.5) for 20 min to quench autofluorescence, blocked with Sea Block Blocking Buffer (Thermo Fisher Scientific) for 40 min, Fc Receptor Blocker (Innovex Biosciences, Richmond, CA) for 30 min, Mouse Detective (Biocare Medical, Concord, CA) for 30 min, and then followed by incubation with VWF antibody (1:200, Merck-MilliporeSigma ab7356) overnight at 4°C. Sections were then incubated with Alexa Fluor 568 donkey α-rabbit antibody (1:200, Thermo Fisher Scientific) for 1 h, and then incubated with 1 μg/mL Hoechst solution (Sigma Aldrich) to visualize nuclei and mounted with 70% glycerol solution. Five random images per sample were assessed at ×200 magnification. Vessel density is expressed as the average number of VWF-positive vessels (less than 30 μm) per high-power field (HPF). Lung fields containing large vessels or airways were excluded from analysis.

Proliferation of the Pulmonary Vasculature

Immunofluorescence staining of pulmonary tissues for Ki67, a proliferative cell marker, with VE-cadherin, an endothelial cell (EC) marker, was performed to assess proliferative pulmonary vascular EC in peripheral pulmonary vessels (less than 100 μm); n = 3 per genotype and condition. The lung sections from control and SM22α-PHD1/2^−/− mice were treated as described earlier, with the following exceptions: samples were incubated with Ki67 (1:200, Abcam ab16667) and VE-cadherin (1:50, Santa Cruz Biotechnology sc-9989) antibodies overnight at 4°C. Sections were then incubated with Alexa Fluor 568 donkey α-rabbit antibody (1:200, Thermo Fisher Scientific) and Alexa Fluor 488 goat α-mouse antibody (1:200, Thermo Fisher Scientific) for 1 h, and then incubated with 1 μg/mL Hoechst solution (Sigma Aldrich) to visualize nuclei and mounted with 70% glycerol solution. Ten random images per sample were assessed at ×400 magnification. Results represent the average number of Ki67-positive cells per VE-cadherin-positive vessel (%).

Assessment of Collagen IV Expression

Immunofluorescence staining for α-SMA, an SMC marker, and collagen IV, a vascular extracellular matrix (ECM) marker was performed to visualize the basement membrane of peripheral pulmonary vessels (less than 100 μm); n = 3 mice per genotype per condition. The lung sections from control and SM22α-PHD1/2^−/− mice were treated as described earlier, with the following exceptions: samples were incubated with Collagen IV (1:100, MilliporeSigma AB756P) and α-SMA (1:200, Abcam ab5694) antibodies overnight at 4°C. Ten random images per sample were assessed at ×400 magnification. Results shown represent the average number of collagen IV-positive cells per α-SMA-positive vessel (%).

RNAscope In Situ Hybridization

P14 murine lungs were inflated by 2% agarose solution through trachea cannulation, and then fixed in 10% neutral buffered formalin for 16 h. Probes for mouse Tagln (SM22α), Cdh5 (VE-cadherin), and Angpt2 (angiopoietin-2) were designed by Advanced Cell Diagnostics USA (ACD, Newark, CA). In situ validation was performed using the RNAscope Multiplex Fluorescent v2 Assay kit (ACD) according to the manufacturer’s protocol. Formalin-fixed paraffin-embedded (FFPE) lung sections (5 μm) were used within a day of sectioning for optimal results. Nuclei were counterstained with DAPI (Life Technology Corp.). Opal dyes (Akoya Biosciences) were used for signal amplification as directed by the manufacturer. According to the manufacture’s RNAscope protocol (RNAscope Multiplex Fluorescent Reagent kit v2 Assay, ACD), fixed lung tissues were paraffin-embedded and 5 μm sections were processed. Lung tissue sections were dried overnight, and deparaffinized with xylene and 100% ethanol. Then, the slides were covered with hydrogen peroxide for 10 min at room temperature and washed in distilled water twice. The slides were incubated in a steamer with distilled water for 10 s and then incubated with RNAscope 1× Target Retrieval Reagent for 15 min. Then the slides were washed in distilled water for 15 s at room temperature, transferred into 100% ethanol for 3 min, and dried at 60°C for 5 min. Tissue sections were covered with the appropriate probe mix and incubated in the HybEZ Oven at 40°C for 2 h. The signal was amplified by subsequent incubation of Amp-1, Amp-2, and Amp-3, one step each for 30, 30, and 15 min, respectively, at 40°C using the HybEZ Oven. Each incubation step was followed by a 2 min wash twice using RNAscope washing buffer. Each probe was amplified by incubation with horseradish peroxidase (HRP), followed by the addition of the appropriate fluorophore diluted with the ACD-provided TSA buffer. Each channel was then blocked using an HRP blocker before amplification of the next channel. Nucleic acids were stained using the manufacturer’s supplied DAPI (ACD) for 30 s and washed twice with PBS. Images were captured with Zeiss LSM 780 confocal microscopes (Carl Zeiss Microscopy GmbH, Gottingen, Germany) at ×630 magnification, using 405 nm, 488 nm, 560 nm, and 633 nm excitation lasers. For scanning tissue, each image frame was set as 1,024 × 1,024 and pinhole 1 Airy Unit (AU). For providing Z-stack confocal images, the Z-stack panel was used to set z-boundary and optimal intervals, and images with maximum intensity were processed by merging Z-stack images. For all, both merged signal and split channels were collected. To analyze the images, ×630 magnification images were used for regional analyses by counting the number of Angpt2 signals and Cdh5-positive cells in the images. Separate images (5–8) were used for each animal (n = 4–6 per genotype and condition) and data are represented as the average total count of signals in Tagln-positive cells per vessel, and the average total number of Cdh5-positive cells per HPF.

Tube Forming Assay

To analyze tube forming function, isolated 100,000 PEC were seeded onto 300 μL of Growth Factor Reduced (GFR) Matrigel Matrix Basement Membrane (BD) coating collagen-coated coverslips per 35-mm plate and incubated with 100,000 PHD1/2^+/+ or PHD1/2^−/− PASMC. In separate experiments, cell identity was confirmed after 2 days of coculture via staining with α-SMA (1:400, Sigma Aldrich A2547) and VE-cadherin (1:50, Santa Cruz Biotechnology sc-9989) antibodies. Plates were incubated in starvation media (0.1% FBS SMC media) or 0.1% FBS SMC media + 220 ng/mL angiopoietin-2 (ANG2; Recombinant mouse Angiopoietin 2, R&D Systems). For visualization of vascular structures within the Matrigel Matrix, the following method was used (25): After 5 days of incubation, samples were fixed in 2% paraformaldehyde for 45 min at 37°C, incubated with IF buffer (130 mM NaCl, 7 mM Na₂HPO₄, 3.5 mM NaH₂PO₄, 7.7 mM NaN₃, 0.1% BSA, 0.2% Triton X-100, 0.05% Tween 20, 10% goat serum) for 2 h at 37°C to permeabilize and block vascular networks, incubated overnight with phalloidin-488 rhodamine (PHDR1, Cytoskeleton) in IF buffer at 37°C to recognize f-actin, incubated with 1 μg/mL Hoechst solution (Sigma) for 15 min at 37°C to visualize nuclei, and then mounted with Permount solution (Fisher Scientific). Vascular network formations were assessed at ×200 magnification. Tube length, number of tubes per colony, and number of branches per tube were quantified at ×200 magnification with 30 randomized images examined per sample.

Statistical Analysis

All experiments were performed a minimum of three times, unless otherwise indicated. Statistical analysis was performed with the GraphPad Prism 9.0 software package (GraphPad Software). Results are presented as means ± SE. Statistical differences between two groups were determined by Student’s t test and between more than two groups and conditions (normoxia and hyperoxia) by two-way ANOVA, followed by Tukey’s multiple comparisons test. In the tube forming assay, one-way ANOVA was used for statistical analysis. A P value of <0.05 was considered statistically significant unless otherwise indicated.

RESULTS

SM22α Cell-Specific Deletion of PHD1 and PHD2 Increased HIF-1α Protein Expression and Nuclear Localization

Deletion of PHD1 and PHD2 was confirmed by analyzing DNA derived from tails from SM22α-PHD1/2 mice. Isolated PASMC from SM22α-PHD1/2 mice were examined by Western immunoblot analysis and immunocytochemistry for gain of HIF-1α expression and nuclear localization (Fig. 1, A–D). PASMC from control mice expressed HIF-1α, but not HIF2 α under normoxic conditions. Compared with controls, PASMC from SM22α-PHD1/2^−/− demonstrated more HIF-1α expression and reduced PHD1 and PHD2 expression under both normoxic and hyperoxic conditions (Fig. 1, A and B). Moreover, in controls 5.1 ± 1.8% of PASMC demonstrated HIF-1α nuclear localization as compared with 13.4 ± 1.2% of PHD1/2^−/− PASMC (Fig. 1, C and D).

Figure 1.

SM22α cell-specific deletion of prolyl hydroxylase (PHD)1 and PHD2 increases pulmonary artery smooth muscle cells (PASMCs) hypoxia-inducible factor (HIF)-1α protein expression and nuclear localization. A: immunoblot detection of HIF-1α, PHD1, and PHD2 proteins in isolated PASMC under normoxic and hyperoxic conditions. B: quantification of HIF-1α, PHD1, and PHD2 protein expression in isolated PASMC. C: representative images of HIF-1α expression in isolated PASMC. Cytoplasmic HIF-1α, red; nuclear HIF-1α, magenta; nuclei, blue; magnification ×200. D: quantification of nuclear HIF-1α expression in isolated PASMC. Data are presented as means ± SE; *P < 0.05, ***P < 0.001; n = 5/genotype/condition.

SM22α Cell-Specific HIF-1α Stabilization Mitigated Hyperoxia-Induced Neonatal Lung Injury

In control mice maintained in normoxia through 14 days of life, the radial alveolar count (RAC) and mean linear intercept (MLI) were 9.0 ± 0.1 and 18.0 ± 0.4 μm, respectively. In control mice exposed to hyperoxia for 14 days, RAC decreased to 4.6 ± 0.2 (P < 0.0001, vs. normoxia) and MLI increased to 33.5 ± 0.7 μm (P < 0.0001, vs. normoxia). Under normoxic conditions, RAC and MLI in lungs from SM22α-PHD1/2^−/− mice did not differ from control mice. Hyperoxic lung injury was attenuated in SM22α-PHD1/2^−/− mice, compared with controls, with a RAC of 6.8 ± 0.2 (P < 0.0001, vs. control hyperoxia) and an MLI of 25.8 ± 0.9 μm (P < 0.0001, vs. control hyperoxia) (Fig. 2, A and B).

Figure 2.

SM22α cell-specific hypoxia-inducible factor (HIF)-1α stabilization mitigates the effects of hyperoxia-induced neonatal lung injury. A: representative hematoxylin and eosin (H&E)-stained sections of pulmonary tissues from postnatal day 14 (P14) SM22α-prolyl hydroxylase (PHD)1/2 mice neonatally exposed to normoxia or hyperoxia for 14 days. Scale bar = 100 μm. B: radial alveolar counts (RACs) of P14 SM22α-PHD1/2 mice. Scale bar = 100 μm. C: mean linear intercepts (MLI) of P14 SM22α-PHD1/2 mice. Scale bar = 100 μm. D: representative images of von Willebrand factor (VWF)-positive pulmonary vessels in P14 SM22α-PHD1/2 mice. VWF, red; nuclei, blue; scale bar = 100 μm. E: quantification of microvessel density (vessels <30 μm) in P14 SM22α-PHD1/2 mice. Data are presented as means ± SE; **P < 0.01, ***P < 0.001 (SM22α-PHD1/2^−/− vs. SM22α-PHD1/2^+/+); §§P < 0.01, §§§P < 0.001 (Hyperoxia vs. Normoxia); n = 6–12/genotype/condition.

Interestingly, at 14 days of normoxia, mean septal thickness was increased in SM22α-PHD1/2^−/− mice compared with control mice (7.5 ± 0.3 μm vs. 5.8 ± 0.1 μm, P < 0.05). Hyperoxia exposure increased septal thickness in control mice (8.4 ± 0.4 μm, P < 0.001, vs. normoxia), but not in SM22α-PHD1/2^−/− mice (8.4 ± 0.5 μm vs. 7.5 ± 0.3 μm, P = NS). Hyperoxia exposure decreased vessel density in both genotypes. In control mice, the number of VWF-positive vessels per high power field (HPF) decreased from 7.9 ± 1.1 to 3.9 ± 0.4 (P < 0.01) with hyperoxia. In SM22α-PHD1/2^−/− mice, the number of VWF-positive vessels per high power field (HPF) decreased in hyperoxia (11.9 ± 0.5, normoxia) versus 5.8 ± 0.5, hyperoxia (P < 0.0001), but to a lesser degree relative to the control mice. SM22α-PHD1/2^−/− mice demonstrated significantly more VWF-positive vessels compared with control mice under both normoxic and hyperoxic conditions (P < 0.05; Fig. 2, D and E).

SM22α Cell-Specific HIF-1α Stabilization Mitigated the Long-Term Effects of Hyperoxia-Induced Lung Injury

To assess the long-term effects of neonatal hyperoxia exposure, mice exposed to hyperoxia for 14 days were maintained in normoxia until 12 wk of age [Hyperoxia-recovery (R)]. In control mice (recovery), lung histology improved over time with an increase in RAC, and decreases in MLI, and septal thickness. In PHD1/2^−/− mice exposed to hyperoxia, RAC increased from 6.2 ± 0.2 (14 days) to 7.4 ± 0.2 at 12 wk (P < 0.05) and MLI decreased from 35.8 ± 1.8 μm to 31.2 ± 1.0 μm (P < 0.05) compared with control mice (Fig. 3, A and B). At 12 wk of age, pulmonary microvascular density was greater in SM22α-PHD1/2^−/− compared with controls in both normoxia (9.1 ± 1.0 vs. 5.8 ± 0.1, P < 0.01) and in Hyperoxia-R (7.6 ± 0.7 vs. 3.3 ± 0.1, P < 0.001). In SM22α-PHD1/2^−/− mice, microvessel density was similar in both normoxia and Hyperoxia-R. In contrast, microvascular density was decreased in Hyperoxia-R mice, compared with age-matched normoxia mice (Fig. 3, D and E; P < 0.05).

Figure 3.

SM22α cell-specific hypoxia-inducible factor (HIF)-1α stabilization mitigates the long-term effects of neonatal lung injury. A: hematoxylin and eosin (H&E)-stained sections of pulmonary tissues from adult SM22α-prolyl hydroxylase (PHD)1/2 mice (12 wk) neonatally exposed to normoxia or hyperoxia (Hyperoxia Recovery) for 14 days. SM22α-PHD1/2 mice exposed to normoxia for 12 wk served as controls. Scale bar = 100 µm. B: radial alveolar counts (RAC) of adult SM22α-PHD1/2 mice. Scale bar = 100 μm. C: mean linear intercepts (MLI) of adult SM22α-PHD1/2 mice. Scale bar = 100 μm. D: representative images of von Willebrand factor (VWF)-positive pulmonary vessels in adult SM22α-PHD1/2 mice (12 wk). VWF, red; nuclei, blue; scale bar: 100 μm. E: quantification of microvessel density (vessels <30 μm) in adult SM22α-PHD1/2 mice. Data are presented as means ± SE; *P < 0.05, **P < 0.01, ***P < 0.001 (SM22α-PHD1/2^−/− vs. SM22α-PHD1/2^+/+); §P < 0.05, §§§P < 0.001 (Hyperoxia vs. Normoxia); n = 5–8/genotype/condition.

SM22α Cell-Specific HIF-1α Stabilization Increased Pulmonary Endothelial Cell Proliferation and Collagen IV

Given the differences in microvascular density between the genotypes, we measured Ki67, expression, a marker for cell proliferation. Under both normoxic and hyperoxic conditions, Ki67 expression was increased in VE-cadherin+ cells in the SM22α-PHD1/2^−/−, compared with control, mice condition (4.7 ± 0.1 vs. 1.6 ± 0.5, P < 0.01 in normoxia; 5.8 ± 0.2 vs. 2.1 ± 0.7, P < 0.01 in hyperoxia) (Fig. 4, A and B). Interestingly, hyperoxia did not decrease Ki67 expression in VE-cadherin+ cells in either genotype. To further examine the changes in the vasculature in the SM22α-PHD1/2^−/− mice, collagen IV expression was quantified in vessels less than 100 μm. Collagen IV expression decreased from 41.5 ± 5.9% in normoxia to 10.0 ± 2.0%, P < 0.01, in control mice. However, in SM22α-PHD1/2^−/− mice, hyperoxia had no effect on collagen IV expression (59.4 ± 2.2% vs. 57.6 ± 5.8%, P = NS) (Fig. 4, C and D).

Figure 4.

SM22α cell-specific hypoxia-inducible factor (HIF)-1α stabilization increases pulmonary endothelial cell proliferation and collagen IV content. A: representative images of VE-cadherin- and Ki67-positive pulmonary vessels in postnatal day 14 (P14) SM22α-prolyl hydroxylase (PHD)1/2 mice neonatally exposed to normoxia or hyperoxia for 14 days. VE-cadherin, green; Ki67, red; nuclei, blue; scale bar: 50 μm. B: quantification of Ki67-positive cells in VE-cadherin-positive vessels from SM22α-PHD1/2 mice. C: representative images of α-smooth muscle actin (α-SMA)- and collagen IV (Col IV)-positive pulmonary vessels in P14 SM22α-PHD1/2 mice. α-SMA, green; Col IV, red; nuclei, blue; scale bar = 50 μm. D: quantification of cells positive for Col IV in P14 SM22α-PHD1/2 mice. Data are presented as means ± SE; **P < 0.01, ***P < 0.001 (SM22α-PHD1/2^−/− vs. SM22α-PHD1/2^+/+); §§P < 0.01 (Hyperoxia vs. Normoxia); n = 3/genotype/condition.

SM22α Cell-Specific HIF-1α Stabilization Increased Angiopoietin-2 Expression

To determine whether specific HIF-1α downstream targets account for the relatively well-preserved pulmonary microvasculature in the SM22α-PHD1/2 mice, lungs were evaluated using RNAscope to determine Angpt2, Tagln, and Cdh5 expression. In neonatal SM22α-PHD1/2^−/− mice, compared with control mice, Angpt2 expression was increased in Tagln-expressing cells in both normoxia (2.6 ± 0.3 vs. 1.0 ± 0.2, P < 0.01) and hyperoxia (5.4 ± 0.3 vs. 2.0 ± 0.2, P < 0.0001), Moreover, hyperoxia increased Angpt2 expression in Tagln-expressing cells in both genotypes (P < 0.05 in control mice, P < 0.001 in SM22α-PHD1/2^−/− mice). Furthermore, in SM22α-PHD1/2^−/− mice, Cdh5+ cells were increased in SM22α-PHD1/2^−/− compared with control mice in both normoxia (32.0 ± 1.2 vs. 23.4 ± 1.1, P < 0.001) and hyperoxia (25.7 ± 1.7 vs. 17.8 ± 0.6, P < 0.01). Hyperoxia decreased Cdh5+ expression in both genotypes (P < 0.05 in control mice, P < 0.01 in SM22α-PHD1/2^−/− mice; Fig. 5, A–C).

Figure 5.

SM22α cell-specific hypoxia-inducible factor (HIF)-1α stabilization increases angiopoietin-2 expression. A: representative images of pulmonary tissues from postnatal day 14 (P14) SM22α-PHD1/2 mice exposed to normoxia or hyperoxia for 14 days using RNAscope in situ hybridization. Cdh5 as a probe for VE-cadherin mRNA, green; Tagln, SM22α, yellow; Angpt2, angiopoietin-2, red; DAPI; scale bar = 20 μm. B: quantification of angiopoietin-2 signals in SM22α-expressing cells of the pulmonary arterioles from SM22α-PHD1/2 mice. C: quantification of Cdh5-positive cells in pulmonary tissues from SM22α-PHD1/2 mice. Data are presented as means ± SE; **P < 0.01, ***P < 0.001 (SM22α-PHD1/2^−/− vs. SM22α-PHD1/2^+/+); §P < 0.05, §§P < 0.01 §§§P < 0.001 (Hyperoxia vs. Normoxia); n = 4–6/genotype/condition.

SM22α Cell-Specific HIF-1α Stabilization Increased Angiogenesis

To address the potential that secreted factors from PASMC that are PHD1/2^−/−might be modulating angiogenesis, coculture experiments were performed. Tube formation assays were performed with P14 PEC and adult PASMC from either PHD1/2^+/+ or PHD1/2^−/− mice (Fig. 6A). PEC alone formed branching tubes, and coculture of PEC with PHD1/2^−/− compared with PHD1/2^+/+ significantly increased the number of tubes per colony, tube length, and branching. Coculture of PEC with PHD1/2^−/−, compared with PHD1/2^+/+ led to more tubes per colony, increased tube length, and more branching. The addition of recombinant Ang2 protein to PHD1/2^+/+ PASMC cocultured with PEC increased tubes per colony (5.7 ± 0.5 vs. 3.5 ± 0.3, P < 0.001) and branching (7.4 ± 0.6 vs. 4.5 ± 0.3 branches per colony, P < 0.001) than cocultures of PHD1/2^+/+ PASMC and PEC alone. Ang2-supplementation in PHD1/2^−/− PASMC co-cultured with PEC increased tube formation (6.0 ± 0.4 vs. 4.3 ± 0.3, P < 0.01), branching (9.8 ± 0.5 vs. 6.4 ± 0.5, branches per colony, P < 0.001), and tube length (109 ± 3.5 μm vs. 69.8 ± 2.8 μm, P < 0.001) than cocultures of PHD1/2^−/− PASMC and PEC alone (Fig. 6, B–D).

Figure 6.

SM22α cell-specific hypoxia-inducible factor (HIF)-1α stabilization increases angiogenesis in vitro. A: representative images from tube forming assays using pulmonary endothelial cells (PECs) from postnatal day 14 (P14) control mice and pulmonary artery smooth muscle cells (PASMCs) from adult SM22α-prolyl hydroxylase (PHD)1/2 mice. Phalloidin, green; nuclei, blue; scale bar = 100 μm. B: quantification of number of tubes per colony of PEC cocultured with PASMC from SM22α-PHD1/2 mice. C: quantification of the number of branch outgrowths per tube. D: quantification of tube length. Data are presented as means ± SE; *P, vs. PEC: *P < 0.05, **P < 0.01, ***P < 0.001; +P value between genotypes: +P < 0.05, ++P < 0.01, +++P < 0.001; §P, between PEC/PASMC and PEC/PASMC+Ang2: §§P < 0.01, §§§P < 0.001; n = 5 animals/genotype, 5-8 separate experiments/condition, and 30 images/group. E: representative brightfield images from tube forming assays. Scale bar = 100 µm.

DISCUSSION

To address the cell-specific role of HIF in hyperoxia-induced neonatal lung injury activity, PHD1 and PHD2, two hydroxylases responsible for the turnover of HIF, were deleted from SM22α-expressing cells. This strategy led primarily to increased expression of HIF-1α, but not HIF-2α, in SM22α-expressing cells. In normoxia, alveolarization was similar in SM22α-PHD1/2^−/− mice and controls. The deleterious effect of neonatal hyperoxia on lung structure was significantly more pronounced in control, compared with SM22α-PHD1/2^−/− mice, underscoring a cell-specific, protective role for HIF-1α. Moreover, HIF-1α stabilization in SM-22α cells led to durable improvement in lung structure that persisted through adulthood in SM22α-PHD1/2^−/− mice following recovery from neonatal hyperoxic exposure relative to control mice. Consistent with the vascular hypothesis of neonatal lung development, SM22α-HIF-1α stabilization increased pulmonary vascular density under normoxic conditions and preserved vascular density, and alveolarization, even with exposure to hyperoxia. Furthermore, expression of angiopoietin-2, a HIF-1α target, was greater in SM22α-PHD1/2^−/− mice, compared with controls, and ANG2 supplementation restored the capacity of SM22α-expressing cells to promote angiogenesis in vitro. Taken together, these observations demonstrate a previously unknown, proangiogenic role for HIF-1α in SM22α-expressing cells that mitigate lung injury even in the context of prolonged hyperoxia exposure.

These results point to cell-specific role for HIF-1α in SM22α-expressing cells. Though loss of PHD1 and 2, albeit with a small amount of residual expression owing perhaps to insufficient Cre expression to completely delete PHD expression (26), stabilizes all HIF isoforms (27, 28), HIF-1α predominates in SM22α-expressing cells(29). HIF-1α and HIF-2α possess distinct expression patterns in the lung suggesting isoform- and cell-specific functions (30). Genetic deletion of HIF-1α is embryonically lethal (15), and deletion of HIF-2α leads to fetal death in ∼50% of embryos and compromises lung development in surviving pups, resulting in neonatal respiratory distress and death (31). HIF stabilization via pharmacologic, noncell specific, inhibition of PHDs preserves lung structure and function in the setting of injuries such as hyperoxia (32) and mechanical ventilation (17), and mitigates the adverse effects of intrauterine inflammation on lung growth (18). The observation that HIF-1α stabilization in the pulmonary epithelium disrupts branching morphogenesis, impairs lung maturation, and causes pulmonary hemorrhage underscores the importance of cell- and isoform-specific knowledge (33). In the lung, HIF possesses cell- and isoform-specific roles into adulthood as HIF stabilization in pulmonary artery endothelial cells leads to pulmonary hypertension and vascular remodeling due to increased HIF-2α, and not HIF-1α (34, 35). To our knowledge, these results are the first to outline a cell-specific role for HIF-1α produced by SM22α-expressing cells in mitigating hyperoxia-induced neonatal lung injury.

Since the pulmonary endothelial cell (PEC) is the primary driver of angiogenesis in the developing lung (5), we postulated that a secreted factor from SM22α-expressing cells might account for the increased vascular density and protection from hyperoxic lung injury. Using coculture experiments, we demonstrated that in the presence of PASMC with HIF stabilization, pulmonary artery endothelial cells formed more and longer tubes. Moreover, we noted a marked increase in Ang2 expression in HIF-1α stabilized PASMC. Even under hyperoxic conditions, SM22α-expressing cells secreted ANG2 and augmented PEC tube formation. These results outline a noncell-autonomous role for HIF-1α in supporting PEC angiogenesis via ANG2 secretion from SM22α-expressing cells. Indeed, results in SM22α−PHD1/2^−/− mice, compared with controls, wherein Ang2 expression was increased in Tgln-expressing cells under conditions of both normoxia and hyperoxia, as well as increased Cdh5 expression was consistent with the in vitro studies.

Taken together these observations support a proangiogenic role for ANG2 in the present model. Although ANG1 is an agonist for the Tyr kinase with Ig and epidermal growth factor homology domains (Tie2), ANG2 is a context-dependent antagonist that can inhibit ANG1-induced Tie2 phosphorylation (36). Though the role of ANG2 has been characterized as antiangiogenic and is increased in the bronchoalveolar lavage fluid of infants with BPD (37), suggesting a deleterious role for the molecule, there is increasing appreciation of the complexity and context-specific effects of ANG2 (38). When ANG2 competes for receptor binding in the presence of VEGF, endothelial junctional integrity is compromised, pericyte-endothelial cell interactions are decreased, and EC proliferation, migration, and vascular sprouting are increased, alterations that favor angiogenesis. In contrast, in the absence of VEGF, ANG2 causes EC apoptosis (39). In the presence of constitutive HIF stabilization in SM22α-expressing cells, levels of VEGF and ANG2, downstream HIF targets, are likely increased that may underlie the increased angiogenesis and more well-preserved lung structure after hyperoxia exposure in SM22α−PHD1/2^−/− mice, compared with controls.

Moreover, this study describes a potential role for HIF-1α produced by SM22α-expressing cells in maintaining expression of collagen IV. In SM22α-PHD1/2^−/− mice, there was more collagen IV expression under both normoxic and hyperoxic conditions than in control mice. Collagen IV is an essential component of the basement membrane in the lung parenchyma that separates the epithelial, mesenchymal, and endothelial compartments of the lung (40). Formation of the basement membrane is a critical determinant of secondary septation of the lung (41). Alterations in basement membrane composition, as evident in the control mice, but not in SM22α−PHD1/2^−/− mice, are present in multiple pediatric lung diseases, including bronchopulmonary dysplasia, the disease that neonatal exposure to hyperoxia is designed to model (4). In human infants, collagen IV is increased in bronchoalveolar lavage fluid of infants with BPD, compared with age-matched controls underscoring an association between collagen IV destruction and BPD (42). In the context of hyperoxia exposure, preserved collagen IV can promote angiogenesis and assist in maintaining epithelial integrity (43). In contrast, loss of collagen IV compromises new vessel formation and epithelial integrity (43). How HIF stabilization results in more and perhaps better-organized collagen IV merits further study.

Hypoxia-inducible factors (HIFs), O₂-sensitive transcription factors, are essential regulators of angiogenesis (44), with downstream effects that are both temporal and cell-specific (18, 32, 45, 46). The present data demonstrate that HIF activity in select subsets of lung mesenchymal cells plays a key role in postnatal angiogenesis and alveolarization, even in the context of neonatal lung injury due to hyperoxia exposure. In these genetic gain-of-function experiments, we used a Cre-recombinase driven by the Tagln promoter. However, the gene is expressed by multiple mesenchymal cell types including vascular smooth muscle cells, airway smooth muscle cells, myofibroblasts, and pericytes (47). Thus, our present findings demonstrating that HIF stabilization induced by deleting prolyl hydroxlases (PHD)1 and 2 do not permit us to definitively state which subtypes of the Tagln-expressing mesenchymal cell types are playing the most central role (27) in preserving pulmonary vascular and alveolar growth in neonatal mice exposed to hyperoxia. Further temporal and cell-specific resolution can be achieved by using the same strategy of PHD1 and 2 deletion using an inducible Cre with a more specific driver such as Notch3 to target pericytes and vascular smooth muscle cells specifically (48). Another limitation of the present work is that deletion of PHD1 and 2 increased the expression of all HIF isoforms. Although we demonstrated increased expression of HIF-1α primarily, there is potential HIF-2α exerted a disproportionately significant effect. A strategy that employs deletion of either Hifa or Epas1 would permit definitive identification of the isoform that confers protection from hyperoxia-induced lung injury.

Overall, the present work demonstrates a cell-specific role for HIF in protecting the developing lung from hyperoxia-induced lung injury. The protective effects of cell-specific increases in HIF are durable given the relatively well-preserved lung structure in adult mice. These data point to a noncell-autonomous effect of HIF-1a as stabilization in the SM22a cells preserves endothelial proliferation, microvascular density, collagen IV content, and lung structure, perhaps via increased production of the HIF targets ANG2 and VEGF. These data underscore the importance of insight into the cell- and isoform-specific role of HIF, a transcription factor central to all organismal responses to low oxygen tension states and incompletely understood roles.

DATA AVAILABILITY

The data that support the findings of this study will be made available upon reasonable request from the corresponding author.

GRANTS

This work has been supported by the National Institutes of Health Grants HL-060784 (to D. N. Cornfield) and HL-160018 (to D. N. Cornfield and C. M. Alvira).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.I., E.A.B., C.M.A., and D.N.C. conceived and designed research; R.I., E.A.B., and X.C. performed experiments; R.I., E.A.B., X.C., and D.N.C. analyzed data; R.I., E.A.B., X.C., C.M.A., and D.N.C. interpreted results of experiments; R.I. and E.A.B. prepared figures; R.I., E.A.B., and D.N.C. drafted manuscript; R.I., E.A.B., C.M.A., and D.N.C. edited and revised manuscript; R.I., E.A.B., C.M.A., and D.N.C. approved final version of manuscript.