Endothelial Adenosine Monophosphate-Activated Protein Kinase-Alpha1 Deficiency Potentiates Hyperoxia-Induced Experimental Bronchopulmonary Dysplasia and Pulmonary Hypertension

Bronchopulmonary dysplasia and pulmonary hypertension, or BPD-PH, are serious chronic lung disorders of prematurity, without curative therapies. Hyperoxia, a known causative factor of BPD-PH, activates adenosine monophosphate-activated protein kinase (AMPK) α1 in neonatal murine lungs; however, whether this phenomenon potentiates or mitigates lung injury is unclear. Thus, we hypothesized that (1) endothelial AMPKα1 is necessary to protect neonatal mice against hyperoxia-induced BPD-PH, and (2) AMPKα1 knockdown decreases angiogenesis in hyperoxia-exposed neonatal human pulmonary microvascular endothelial cells (HPMECs). We performed lung morphometric and echocardiographic studies on postnatal day (P) 28 on endothelial AMPKα1-sufficient and -deficient mice exposed to 21% O2 (normoxia) or 70% O2 (hyperoxia) from P1–P14. We also performed tubule formation assays on control- or AMPKα1-siRNA transfected HPMECs, exposed to 21% O2 or 70% O2 for 48 h. Hyperoxia-mediated alveolar and pulmonary vascular simplification, pulmonary vascular remodeling, and PH were significantly amplified in endothelial AMPKα1-deficient mice. AMPKα1 siRNA knocked down AMPKα1 expression in HPMECs, and decreased their ability to form tubules in normoxia and hyperoxia. Furthermore, AMPKα1 knockdown decreased proliferating cell nuclear antigen expression in hyperoxic conditions. Our results indicate that AMPKα1 is required to reduce hyperoxia-induced BPD-PH burden in neonatal mice, and promotes angiogenesis in HPMECs to limit lung injury.


Introduction
Bronchopulmonary dysplasia (BPD) remains the most common adverse outcome in preterm neonates, and is still one of the most challenging complications in perinatal medicine. With the improved survival of extremely premature infants, the incidence of BPD remains high, at about 30%, depending on the cohort and the definition used [1][2][3][4]. Pulmonary hypertension (PH) is one of the most serious long-term morbidities of BPD. PH
Collagenase digestion was performed for 30 min at 37 • C on a rotary agitator, at a speed of 125 RPM. Any remaining undigested tissue was mechanically disrupted by passing through an 18 G needle attached to a 5-mL syringe three times, followed by passage through a 70 mm cell strainer. The cells were then centrifuged at 400 g for 5 min at 4 • C. The supernatant was discarded, and the red blood cells (RBCs) in the cell pellet were lysed using RBC lysis buffer, following which, the cell pellet was harvested by centrifugation at 400 g for 5 min at 4 • C, and discarding the supernatant. The cell pellet was resuspended in PBS containing 0.1% BSA, and incubated with anti-platelet and endothelial cell adhesion molecule 1 (PECAM-1) antibody-conjugated Dynabeads from Life Technologies (Carlsbad, CA, USA) on a rocker for 30 min at room temperature, as per the manufacturer's instructions. After isolation, cells were washed with PBS three times, and the protein was extracted using a lysis buffer (Santa Cruz Biotechnologies, Santa Cruz, CA, USA; sc-24948), as per the manufacturer's recommendations. The lung endothelial cell protein lysates were then subjected to immunoblotting using antibodies against: AMPKα1 (Abcam, Cambridge, UK; ab3759), AMPKα2 (Abcam; ab3760), and glyceraldehyde 3-phosphate dehydrogenase ([GAPDH] Cell Signaling, Danvers, MA, USA; 2118).

Analysis of Alveolarization and Pulmonary Vascularization
The alveolarization was evaluated on P28 by quantifying mean linear intercepts (MLI) and the radial alveolar counts (RAC), as described before [26]. Lung vascular development was also determined on P28 by quantifying vWF-stained lung blood vessels with a diameter of <150 µm [31].

Pulmonary Vascular Remodeling
Pulmonary vascular remodeling was evaluated by quantifying the medial thickness index of resistance pulmonary blood vessels. Deparaffinized lung tissues were subjected to immunostaining using α-smooth muscle actin (α-SMA) antibody (Sigma-Aldrich, St. Louis, MO, USA; A5228), and the medial thickness index was estimated using the equation: [(areaext − areaint)/areaext] × 100, where areaext and areaint represent the areas within the external and internal borders of the α-SMA layer, respectively [23].

Echocardiography
Transthoracic echocardiography was performed on P28, to evaluate the indices of PH, as described previously [26]. Pulsed-wave Doppler recording of the pulmonary blood flow obtained at the aortic valve level in the parasternal right ventricular outflow view was used to estimate pulmonary acceleration time (PAT) [33]. The right ventricular systolic pressure (RVSP) was estimated by the regression formula RVSP = 63.7 − (1.5 × PAT) [33].

Estimation of the Right Ventricle (RV)/Left Ventricle (LV) Free Wall Thickness Ratio
Hematoxylin and eosin-stained sections of paraffin-embedded heart were used to analyze RV/LV free wall thickness ratio, as we have described recently [34].

Cell Culture
The neonatal human pulmonary microvascular endothelial-like cells (HPMECs) were obtained from the American Type Culture Collection (ATCC ® CRL-3244). We grew these cells based on the manufacturer's protocol, and used cells between passages, five and eight, for our studies. We performed transient transfections with either 50 nM control siRNA (Dharmacon, Lafayette, CO, USA; D-001810) or 50 nM target gene-specific AMPKα1 siRNA (Dharmacon; L-005027), using Lipofectamine RNAiMAX (Life Technologies; 13778030). siRNA-mediated AMPKα1 knockdown was confirmed by RT-PCR analysis and immunoblotting.

Real-Time RT-PCR Assays
We initially checked for the integrity and quality of our RNA by denaturing agarose gel and measuring 260/280 ratio, respectively. Then, we performed real-time quantitative RT-PCR analysis with a 7900HT Real-Time PCR System, using TaqMan gene expression master mix and gene-specific primers (AMPKα1-Hs01562315 and GAPDH-Hs02758991), as described previously [30]. We used GAPDH as the reference gene.

Tubule Formation Assay
We performed a matrigel assay to determine tubule formation [36]. HPMECs transfected with control or AMPKα1 siRNA and exposed to normoxia or hyperoxia for 48 h were loaded onto growth factor-reduced matrigel in a 96-well plate. Tubule formation was quantified 24 h later using the Image J software.

Statistical Analysis
Data analysis was done using GraphPad Prism version 9 software (GraphPad Software, La Jolla, CA, USA), and the results are expressed as means ± SD. p-value of <0.05 was considered significant. In vivo studies: The effects of exposure, gene, and their associated interactions on outcome variables, including pulmonary alveolarization, pulmonary angiogenesis, and indices of PH, were analyzed using analysis of variance (ANOVA). If a statistical significance of either variable or interaction was noted by ANOVA, the post hoc Bonferroni test was performed. In vitro studies: The effects of hyperoxia, AMPKα1 knockdown, and their interactions on tubule formation and PCNA expression were analyzed by ANOVA.

AMPKα1 Deficiency Potentiates Neonatal Hyperoxia-Induced Alveolar and Pulmonary Vascular Simplification in Mice
We identified endothelial AMPKα1-deficient and -sufficient mice by genotyping and immunoblotting analysis. The lung endothelial cell AMPKα1 protein expression was significantly decreased in eAMPKα1 +/− mice compared with eAMPKα1 +/+ mice at P14 ( Figure 1A,B). By contrast, the lung endothelial cell AMPKα2 protein expression was significantly greater in eAMPKα1 +/− mice than in eAMPKα1 +/+ mice at P14 ( Figure 1A,C). P28 mice exposed to 70% O 2 (hyperoxia) from P1 to P14 and allowed to recover in 21% O 2 (normoxia) for 14 days had fewer vWF stained-lung blood vessels than mice who remained in normoxia from P1 to P28 (Figure 2A-E). However, the hyperoxia-induced decrease in vWF stained-lung blood vessels was significantly greater in eAMPKα1 +/− mice (5.53 ± 0.9) than in eAMPKα1 +/+ mice (6.76 ± 0.48) ( Figure 2C-E). Similarly, the lungs of P28 mice exposed to neonatal hyperoxia had fewer and larger alveoli, as evident by decreased RAC and increased MLI, respectively, than the lungs of mice exposed to normoxia ( Figure 3A-F). However, the effects of hyperoxia on RAC and MLI were significantly augmented in eAMPKα1 +/− mice than in eAMPKα1 +/+ mice ( Figure 3C-F). These results indicate that endothelial AMPKα1 deficiency potentiates hyperoxia-induced alveolar and pulmonary vascular simplification.

AMPKα1 Deficiency Potentiates Neonatal Hyperoxia-Induced Experimental PH in Mice
On P28, transthoracic high-resolution echocardiographic studies were performed and indices of PH including PAT and RVSP were estimated to elucidate the effects of the Pulmonary vascular remodeling at postnatal day (P) 28 in endothelial AMPKα1 deficient (eAMPKα1 +/− ) mice exposed to hyperoxia during the first two weeks of life. eAMPKα1 +/+ or eAMPKα1 +/− mice were exposed to either 21% O 2 (normoxia) for 4 weeks, or 70% O 2 (hyperoxia) for 2 weeks, followed by normoxia for 2 weeks, and their lung tissues were collected on P28 for quantifying pulmonary vascular remodeling. (A-D) Representative alpha-smooth muscle actin (α-SMA) stained blood vessels (arrow) from eAMPKα1 +/+ (A,C) or eAMPKα1 +/− (B,D) mice, and exposed to normoxia (A,B) or hyperoxia (C,D). Scale bar = 100 µm. (E) Quantification of pulmonary vascular remodeling by medial thickness index. Values represent the mean ± SD (n = 6-10 mice/group). Significant differences between eAMPKα1 +/+ and eAMPKα1 +/− mice under hyperoxic conditions are indicated by †, p < 0.05. Significant differences between the genotype-matched mice under normoxic and hyperoxic conditions are indicated by *, p < 0.05. (ANOVA: Effect: AMPKα1 and hyperoxia, Interaction: Yes).

AMPKα1 Deficiency Potentiates Neonatal Hyperoxia-Induced Experimental PH in Mice
On P28, transthoracic high-resolution echocardiographic studies were performed, and indices of PH including PAT and RVSP were estimated to elucidate the effects of the AMPKα1 gene, neonatal hyperoxia exposure, and their interactions on PH. Neonatal hyperoxia exposure decreased PAT ( Figure 5A-E) and increased the estimated RVSP ( Figure 5F) compared with mice exposed to normoxia. These effects of hyperoxia were significantly greater in AMPKα1 +/− mice (PAT: 11.67 ± 1.56 ms; RVSP: 46.20 ± 2.34 mmHg) than in AMPKα1 +/+ mice (PAT: 14.84 ± 0.99 ms; RVSP: 41.44 ± 1.48 mmHg), suggesting that AMPKα1 deficiency worsens neonatal hyperoxia-induced PH. The heart rate was comparable among all our experimental groups ( Figure 5G). Next, we quantified right ventricular hypertrophy (RVH), which is a marker of severe PH. Although Fulton's index is used to estimate RVH in rodents, the technical challenges associated with estimating this index in neonatal mice can lead to the inaccurate quantification of RVH. Therefore, we used the RV/LV free wall thickness ratio, an alternatively accepted method of quantifying RVH in neonatal mice. Hyperoxia exposure increased the RV/LV free wall thickness ratio; however, AMPKα1 gene expression did not have an independent effect on RVH in our model ( Figure 6). RV/LV free wall thickness ratio, an alternatively accepted method of quantifying RVH neonatal mice. Hyperoxia exposure increased the RV/LV free wall thickness ratio; how ever, AMPKα1 gene expression did not have an independent effect on RVH in our mod ( Figure 6).

AMPKα1 Signaling Is Necessary for HPMEC Tubule Formation
To determine if AMPKα1 signaling is necessary for lung angiogenesis in human neonates, we used siRNA to knock down AMPKα1 gene in HPMECs. AMPKα1 siRNA decreased the mRNA ( Figure 7A) and protein ( Figure 7B,C) expression of AMPKα1 in both normoxic and hyperoxic conditions. AMPKα1 knockdown decreased HPMEC tubule formation in both normoxic and hyperoxic conditions ( Figure 7D-H). Importantly, knockdown of this gene significantly decreased HPMEC tubule formation in the hyperoxia group (Hyperoxia: SiAMPKα1, 54.83 ± 10.59 vs. SiC, 83.67 ± 9.24; p < 0.05) ( Figure 7F-H). Proliferating cell nuclear antigen (PCNA) plays an important role in cell proliferation, and is often used as an index of cellular proliferation [37]. Whereas hyperoxia exposure did not alter the PCNA protein levels in AMPKα1-sufficient cells in our in vitro model, it significantly reduced the PCNA levels in AMPKα1-deficient cells ( Figure 7I,J), indicating that the AMPKα1 may regulate angiogenesis under hyperoxic conditions, partly via PCNA.

Discussion
In this study, we examined the effects of AMPKα1 knockdown on lung development and PH in a murine BPD model. Furthermore, we performed translational studies using HPMECs to decipher the necessary role of AMPKα1 in lung angiogenesis. We demonstrate that AMPKα1 deficiency potentiates hyperoxia-induced neonatal murine lung injury. Additionally, we show that AMPKα1 signaling is necessary for HPMEC tubule formation.

Discussion
In this study, we examined the effects of AMPKα1 knockdown on lung development and PH in a murine BPD model. Furthermore, we performed translational studies using HPMECs to decipher the necessary role of AMPKα1 in lung angiogenesis. We demon-strate that AMPKα1 deficiency potentiates hyperoxia-induced neonatal murine lung injury. Additionally, we show that AMPKα1 signaling is necessary for HPMEC tubule formation.
Lung vascular health maintains alveolar homeostasis and promotes alveolar growth [38,39]. Impaired angiogenesis disrupts alveolarization [40][41][42]. Additionally, reduced growth capacity and abnormal vasoreactivity and extracellular matrix of the lung endothelial cells increases the BPD-PH risk [43][44][45]. Thus, understanding how the lung vascular system homeostasis or health is maintained is pivotal to provide tailored therapies for BPD-PH in preterm infants. Emerging evidence indicates that AMPK promotes vascular health in several organs, including the lungs [21,[46][47][48][49][50]. Furthermore, we recently observed that hyperoxia increases pulmonary AMPKα activation in a murine model of BPD-PH [19]. Yadav et al. demonstrated that hyperoxia decreased pulmonary AMPK function in rat pups after 10days of hyperoxia, and the decreased p-AMPK levels persisted at P21, 10 days after pups were returned to normoxia [21]. Several other investigators have shown an increase in AMPK activation after shorter exposure to hyperoxia, in human lung fibroblasts and human umbilical vein endothelial cells (HUVECs) [51,52]. However, whether endothelial AMPK, especially the α1 subunit, potentiates or mitigates hyperoxia-induced neonatal lung injury is not well studied, providing a strong premise for our study to clarify its role further.
Tie2-driven Cre recombinase was used to decrease the expression of AMPKα1 in the lung endothelium. Importantly, Tie2 is also expressed in other cell types, including macrophages and monocytes. Thus, it is possible that some of our results reflect the deficiency of AMPKα1 in these hematopoietic cells. However, the endothelial cells are significantly enriched with Tie2, and Tie2-Cre mice are frequently used to study the role of endothelial signaling in lung health and disease [20,53,54]. Furthermore, our hyperoxia model recapitulates the BPD-PH phenotype of infants [26,27]. In alignment with this concept, our hyperoxia-exposed animals displayed alveolar and pulmonary vascular simplification. Significantly, hyperoxia-mediated BPD-PH was potentiated in endothelial AMPKα1-deficient mice. Our findings underpin the lung vascular health's essential role in lung development, i.e., vascular hypothesis [11,55]. Additionally, our results signify the necessary role of endothelial AMPKα1 in mediating lung angiogenesis, and decreasing neonatal lung disease burden when exposed to a risk factor like hyperoxia. AMPKα-deficient mice display increased lung injury when exposed to insults such as lipopolysaccharide [56], particulate matter [57,58], and hemorrhagic shock [59]. Our results add to this existing body of literature, and highlight the protective role of AMPKα in lung injury across the rodent life span. We also demonstrate that the necessary role of AMPKα1 extends beyond murine lungs. Consistent with other studies [15,46,50,60], we show that AMPKα deficiency decreases human lung endothelial cell angiogenesis. PCNA plays a major role in DNA replication, and is frequently used as a marker of cellular proliferation [61]. Thus, our data indicate that AMPKα1 regulates HPMEC tubule formation in hyperoxic conditions partly via PCNA-dependent mechanisms. Interestingly, we noted increased AMPKα2 protein expression in endothelial cell AMPKα1-deficient mice. Pulmonary hypertension is a significant morbidity of endothelial cell AMPKα2 deficient mice [49]. Therefore, the increase in the AMPKα2 protein seen in our model may be a compensatory response to maintain the total endothelial AMPKα function. Nevertheless, the compensatory increase in AMPKα2 protein was insufficient to rescue our AMPKα1-deficient mice from hyperoxiamediated lung injury. We also determined the effects of endothelial AMPKα1 deficiency on pulmonary vascular remodeling and function. Although the gold standard diagnostic test for PH is cardiac catheterization, noninvasive echocardiography has been commonly used to delineate cardiac structure and function in small animals, because of its technical feasibility and reliability for determining experimental PH. Furthermore, the RV systolic time intervals, such as PAT determined by echocardiography, correlate with the PA pressure measured by cardiac catheterization [26,33,62]. PAT correlates inversely, while RVSP correlates directly with the pulmonary artery pressure [62][63][64]. Thus, our findings reinforce the fact that exposure to moderate hyperoxia for a prolonged duration induces PH in neonatal mice. Similarly, hyperoxia induced pulmonary vascular remodeling, another PH biomarker, in our model. Our results also demonstrate that endothelial AMPKα1 deficiency potentiates hyperoxia-induced PH, signifying the necessary role of endothelial AMPKα1 to mitigate experimental PH. Studies in adult rodents have demonstrated the necessary and sufficient function of AMPK in preventing and mitigating PH through several mechanisms, including the inhibition of autophagy and proliferation of pulmonary artery smooth muscle cells [65][66][67]. Our study validates these findings and extends the potential protective effect of endothelial AMPKα1 for BPD-PH. We did not observe a statistically significant independent effect of AMPKα1 on hyperoxia-mediated effect on the RV/LV ratio. One possibility for this observation is that we used mice that were partially deficient in AMPKα1, rather than those that completely lacked AMPKα1.
The strengths of our study include the use of: (1) a rigorous genetic technique determining endothelial AMPKα1 s role in developmental lung injury; (2) high-resolution echocardiography to elucidate endothelial AMPKα1 s effects on cardio-pulmonary function; and (3) human neonatal lung endothelial cells determining AMPKα1 s role in human lung angiogenesis, increasing the clinical relevance of our study. Despite these strengths, our study has a few limitations that we will address later. We did not elucidate the: (1) sexspecific effects of endothelial AMPKα1 deficiency; (2) impact of hyperoxia or endothelial AMPKα1 deficiency on lung function; (3) impact of AMPKα1 activation on hyperoxiainduced experimental BPD-PH; and (4) precise molecular mechanisms through which endothelial AMPKα1 deficiency potentiates neonatal hyperoxic lung injury.
In summary, we show that Tie2-Cre-mediated endothelial AMPKα1 deficiency potentiates hyperoxia-induced experimental BPD-PH in mice. Furthermore, we show that AMPKα1 is required for angiogenesis and PCNA expression in human neonatal lung endothelial cells exposed to hyperoxia, which we speculate are some of the mechanisms through which AMPKα1 regulates experimental BPD-PH (Figure 8). To the best of our knowledge, this is the first study that elucidates the essential role of endothelial AMPKα1 in hyperoxia-mediated experimental BPD-PH, and emphasizes that AMPKα1 is a potential therapeutic target for BPD infants who develop PH. The strengths of our study include the use of: (1) a rigorous genetic technique determining endothelial AMPKα1′s role in developmental lung injury; (2) high-resolution echocardiography to elucidate endothelial AMPKα1′s effects on cardio-pulmonary function; and (3) human neonatal lung endothelial cells determining AMPKα1′s role in human lung angiogenesis, increasing the clinical relevance of our study. Despite these strengths, our study has a few limitations that we will address later. We did not elucidate the: (1) sex-specific effects of endothelial AMPKα1 deficiency; (2) impact of hyperoxia or endothelial AMPKα1 deficiency on lung function; (3) impact of AMPKα1 activation on hyperoxiainduced experimental BPD-PH; and (4) precise molecular mechanisms through which endothelial AMPKα1 deficiency potentiates neonatal hyperoxic lung injury.
In summary, we show that Tie2-Cre-mediated endothelial AMPKα1 deficiency potentiates hyperoxia-induced experimental BPD-PH in mice. Furthermore, we show that AMPKα1 is required for angiogenesis and PCNA expression in human neonatal lung endothelial cells exposed to hyperoxia, which we speculate are some of the mechanisms through which AMPKα1 regulates experimental BPD-PH (Figure 8). To the best of our knowledge, this is the first study that elucidates the essential role of endothelial AMPKα1 in hyperoxia-mediated experimental BPD-PH, and emphasizes that AMPKα1 is a potential therapeutic target for BPD infants who develop PH.

Conclusions
AMPKα1 signaling is necessary to mitigate hyperoxia-induced BPD and PH in neonatal mice and promote angiogenesis in neonatal HPMECs.

Conclusions
AMPKα1 signaling is necessary to mitigate hyperoxia-induced BPD and PH in neonatal mice and promote angiogenesis in neonatal HPMECs.