Mouse lung development and NOX1 induction during hyperoxia are developmentally regulated and mitochondrial ROS dependent

Abstract

Animal models demonstrate that exposure to supraphysiological oxygen during the neonatal period compromises both lung and pulmonary vascular development, resulting in a phenotype comparable to bronchopulmonary dysplasia (BPD). Our prior work in murine models identified postnatal maturation of antioxidant enzyme capacities as well as developmental regulation of mitochondrial oxidative stress in hyperoxia. We hypothesize that consequences of hyperoxia may also be developmentally regulated and mitochondrial reactive oxygen species (ROS) dependent. To determine whether age of exposure impacts the effect of hyperoxia, neonatal mice were placed in 75% oxygen for 72 h at either postnatal day 0 (early postnatal) or day 4 (late postnatal). Mice exposed to early, but not late, postnatal hyperoxia demonstrated decreased alveolarization and septation, increased muscularization of resistance pulmonary arteries, and right ventricular hypertrophy (RVH) compared with normoxic controls. Treatment with a mitochondria-specific antioxidant, (2-(2,2,6,6-tetramethylpiperidin-1-oxyl-4-ylamino)-2-oxoethyl)triphenylphosphonium chloride (mitoTEMPO), during early postnatal hyperoxia protected against compromised alveolarization and RVH. In addition, early, but not late, postnatal hyperoxia resulted in induction of NOX1 expression that was mitochondrial ROS dependent. Because early, but not late, exposure resulted in compromised lung and cardiovascular development, we conclude that the consequences of hyperoxia are developmentally regulated and decrease with age. Attenuated disease in mitoTEMPO-treated mice implicates mitochondrial ROS in the pathophysiology of neonatal hyperoxic lung injury, with potential for amplification of ROS signaling through NOX1 induction. Furthermore, it suggests a potential role for targeted antioxidant therapy in the prevention or treatment of BPD.

despite advances in neonatal care, including routine use of antenatal corticosteroids, exogenous surfactant, gentle ventilation, and improved nutrition, bronchopulmonary dysplasia (BPD) remains the most common morbidity associated with prematurity. Nearly 25% of babies born less than 1,250 g meet clinical criteria for BPD, placing them at increased risk of long-term pulmonary morbidity, neurodevelopmental delay, and death (45). Although early descriptions of BPD identified diffuse fibrosis and inflammation, the postsurfactant era has resulted in a “new BPD,” characterized by arrested lung development with compromised alveolarization and vasculogenesis (27, 36).

Although multiple factors contribute to the development of BPD, including inflammation, infection, and ventilator-associated lung injury, animal models demonstrate that supraphysiological oxygen alone can induce phenotypic changes of BPD (22, 46). Despite known toxicity and guidelines for limited use, therapeutic oxygen remains a common and necessary treatment for respiratory distress syndrome and associated hypoxia in newborns.

Neonatal murine hyperoxia has been used previously to investigate the effects of oxidative stress on lung development and provides insight into mechanisms of BPD. Mouse models approximate the effects of disease in premature infants, as mice are in the saccular stage of lung development at birth and complete alveolarization by 2 wk of age (1). Supraphysiological oxygen during this time results in compromised alveolarization, right ventricular hypertrophy (RVH), and vascular remodeling (4, 6). Furthermore, postnatal (PN) maturation of antioxidant enzyme (AOE) capacities has been demonstrated in the newborn mouse, paralleling AOE deficiencies of premature infants (2, 11, 12, 34). This gradual increase of antioxidant capacities during lung development may help prepare the fetus for the relative hyperoxia of the extrauterine environment. Paradoxically, newborn mice, and neonates of several other species, demonstrate improved survival in hyperoxia compared with adults (3, 23).

We previously demonstrated that despite relative oxygen tolerance in newborn mice, acute hyperoxia induces an exaggerated oxidative stress response in immature, compared with mature, murine pulmonary alveolar cells (12). This response is specific to the mitochondria, as cytosolic oxidative stress does not change during acute exposure (12, 21). Developmental regulation of reactive oxygen species (ROS) generation and PN maturation of AOE suggest that early exposure to hyperoxia may result in greater damage to immature lungs.

To determine whether age at exposure affects the response to hyperoxia, we exposed mice to 72 h of 75% oxygen at birth (early PN hyperoxia) or at day of life 4 (late PN hyperoxia). We hypothesized that a critical early developmental window may exist in which the immature lung is more susceptible to supraphysiological oxygen, resulting in a more severe disruption of lung and pulmonary vascular development compared with that seen during later hyperoxic exposure. Since mitochondria appear to be the early source of ROS in hyperoxia, we further hypothesized that compromised lung development with hyperoxia is mitochondrial ROS dependent and may be attenuated by treatment with a mitochondria-targeted antioxidant.

MATERIALS AND METHODS

Experimental design.

Animal protocols were approved by the Institutional Animal Care and Use Committee at Northwestern University. To determine effect of timing of hyperoxic exposure, neonatal C57BL/6 mice of variable age were maintained in a 75% O₂ chamber (Biospherix, Lacona, NY) or 21% O₂ under standard housing conditions with daily rotation of dams. Exposure to hyperoxia was initiated on PN day 0 (P0-3, or early PN) or PN day 4 (P4-7, or late PN). To determine the significance of timing of hyperoxia on lung, pulmonary vascular, and heart development, mice were exposed to 75% O₂ for 72 h at P0 or P4 and recovered in 21% O₂ until 14 days of life. At that time, lungs were inflation fixed with 4% formalin under 25 cmH₂O pressure, and hearts were removed for analysis. To compare the consequences of early and late PN exposure on protein expression, 75% O₂-exposed and control lungs were analyzed at the completion of 72 h exposure (at P3 and P7 for early and late PN O₂, respectively). These lung samples were PBS perfused and flash frozen for protein analysis. To determine the effect of oxygen concentration, additional experiments were performed exposing mice to 90% O₂ at the same early (P0-3) or late PN (P4-7) time periods with analysis of lung and heart development at 14 days.

For antioxidant experiments, mice were injected daily with a mitochondria-targeted antioxidant, mitoTEMPO [(2-(2,2,6,6-tetramethylpiperidin-1-oxyl-4-ylamino)-2-oxoethyl)triphenylphosphonium chloride; 0.7 μg/g subcutaneous, Enzo Life Sciences, Farmingdale, NY], or vehicle (Dulbecco's phosphate-buffered saline) during the experimental period (P0-3 or P4-7). MitoTEMPO exhibits superoxide and alkyl radical scavenging properties and is targeted to the mitochondria by conjugation to a lipophilic cation (35). Dosage of mitoTEMPO was determined from previously published murine models (32).

Lung and heart analysis.

All lung tissue analysis was performed on blinded samples. For alveolar morphometry, sections (5 μm thick) from inflation-fixed lungs were stained with hematoxylin overnight. Using a microscope equipped with a Pixera digital camera, we acquired 8–10 images from each slide under ×20 magnification. Images were analyzed by a single operator using Scion Image software (Scion, Frederick, MD). Mean alveolar area, septal thickness, and septal count were determined for each image and averaged to yield a composite value for each animal (24). Six to eight images were obtained at ×4 and analyzed for radial alveolar count as originally described by Emery and Mithal (17). Eight to 10 representative images of elastin staining (Verhoeff's) were captured and analyzed for mean septal length (from base of the septum to tip) by use of Aperio Imaging software (Leica Biosystems, Buffalo Grove, IL). Fulton's index, defined as the weight of the right ventricle divided by the combined weight of the left ventricle and the septum [RV/(LV+S)], was determined by microscopic dissection and weighing of fresh heart tissue (4). Dissection was performed by a single operator who was blinded to experimental status of samples. Fulton's index data were compiled from multiple litters for a total of n > 12 for each condition.

Pulmonary vasculature analysis.

To determine average vessel count, immunohistochemistry was performed on 5-μm-thick lung sections using von Willebrand factor primary antibody (1:50; polyclonal rabbit anti-human antibody, Dako, Carpinteria, CA) with rhodamine red anti-rabbit secondary (1:100; Molecular Probes, Carlsbad, CA). Tissue was imaged at ×10 on a Nikon Eclipse TE-300 fluorescence microscope, and eight representative images were acquired from each sample. Pulmonary vessels of ∼50–100 μm in diameter were counted by a single operator blinded to study design. Totals were compiled and averaged for each slide and reported as number of vessels per high-powered field (HPF) of 700 × 450 μm (16).

To determine average medial wall thickness (MWT), hematoxylin and eosin-stained lung sections were imaged at ×40 magnification. Representative (n = 6–8) images of small pulmonary arteries were acquired for each sample and analyzed by use of Image J software (National Institutes of Health, Bethesda, MD). Medial wall thickness was defined as the area of the vessel minus the area of the lumen divided by the total cross-sectional area of the artery (4). Analysis of distal pulmonary arteries (40–80 μm diameter) was performed on α-smooth muscle actin-stained samples as previously described with modification (7); six to eight representative vessels were analyzed from each sample. Muscularization was characterized as minimal (less than 25%), moderate (25–50%), or mostly (>50%).

Protein analysis.

Protein was extracted from PBS-perfused frozen whole lungs by pulverization and suspension in lysis buffer (Millipore, La Jolla, CA) containing protease inhibitor (0.1%, Sigma-Aldrich, St. Louis, MO) and phosphatase inhibitor (1%, Calbiochem, La Jolla, CA). Lysates were separated on SDS-PAGE gels, transferred to nitrocellulose membranes, and blotted with primary antibodies, followed by horseradish peroxidase-conjugated secondary antibodies and ECL reagent (Amersham, GE Life Sciences, Buckinghamshire, UK). Primary antibodies included copper zinc superoxide dismutase (CuZnSOD; rabbit; Enzo Life Sciences, Farmingdale, NY; 1:1,000), NADPH oxidase-1 (NOX1; rabbit; Santa Cruz Biotechnology, Dallas, TX; 1:500), NADPH oxidase-2 (NOX2; rabbit; Santa Cruz Biotechnology, 1:500), NADPH oxidase-4 (NOX4; rabbit; Santa Cruz Biotechnology, 1:500), manganese superoxide dismutase (MnSOD; rabbit; Enzo Life Sciences, 1:2,000), and β-actin (rabbit; Cell Signaling Technology, Danvers, MA; 1:5,000). Membranes were imaged and quantified by using the ChemiDoc XRS+ Molecular Imager and software (Bio-Rad, Hercules, CA). Results were expressed as fold change compared with control.

Statistical analysis.

Data are presented as means ± SE, with replicate measurements of n ≥ 6 for all experiments except where noted. Statistical analysis was carried out by ANOVA followed by post hoc comparison by Tukey-Kramer (GraphPad Software, La Jolla, CA) with P ≤ 0.05 considered to be statistically significant.

RESULTS

Early but not late PN hyperoxia results in compromised alveolarization, decreased septation, increased distal artery muscularization, and RVH.

To determine the significance of timing of hyperoxia exposure on alveolar and septal development, inflation-fixed lung tissue from 14-day-old mice exposed to 75% O₂ at birth (early PN) was compared with those exposed on day of life 4 (late PN) and room air controls (normoxia). Early PN O₂ resulted in increased mean alveolar area (AlvA) compared with normoxic controls (Fig. 1A), which was not observed with late PN O₂ (AlvA of 428.1 ± 12.9, 519.5 ± 29.2, and 429.4 ± 17.1 for normoxia, early, and late PN O₂, respectively). Radial alveolar count was determined as a correlate to alveolar number. These data demonstrated a significant decrease in alveolar count with early but not late PN O₂ (Fig. 1B). However, no difference in alveolar wall thickness was observed with averages of 5.5 ± 0.17, 5.7 ± 0.25, and 5.7 ± 0.44 μm for normoxia, early, and late PN O₂, respectively. Evaluation of septation demonstrated decreased septal count per HPF as well as decreased septal length with early but not late PN O₂ (Fig. 2, A and B).

Fig. 1.

Early but not late postnatal hyperoxia compromises alveolar development. A: 14-day-old mice exposed to 75% O₂ from postnatal days 0–3 (P0-3; early PN) but not P4-7 (late PN) demonstrate increased mean alveolar area compared with normoxic controls (21% O₂). B: 14-day-old mice exposed to early but not late PN O₂ demonstrate decreased radial alveolar count (RAC) compared with normoxic controls. C: representative lung images at ×20 original magnification. D: representative RAC measurements at ×4 original magnification. N, normoxia; E, early PN O₂; L, late PN O₂. *P < 0.05.

Fig. 2.

Early but not late postnatal hyperoxia compromises septation. Fourteen-day-old mice exposed to 75% O₂ from P0-3 (early PN) but not P4-7 (late PN) demonstrate decreased septal count (A) and length (B) compared with normoxic controls (21% O₂). HPF, high-powered field. C: representative septal counts in a portion of a ×20 image. D: representative septal lengths in a portion of a ×40 image. *P < 0.05.

To determine whether timing of PN exposure affects development of RVH, Fulton's index from 14-day-old mice exposed to early or late PN O₂ was compared with normoxic controls. Early but not late PN O₂ resulted in increased RVH (Fulton's index = 0.25 ± 0.01, 0.30 ± 0.01, and 0.26 ± 0.01 for normoxic control, early, and late PN O₂, respectively) (Fig. 3). Although RVH was observed in mice exposed to early PN O₂, neither early nor late PN O₂ resulted in vascular remodeling as determined by average vessel count per HPF (Fig. 4A) or MWT (Fig. 5A). However, α-smooth muscle actin staining demonstrated increased muscularization of distal pulmonary arteries after early but not late PN O₂ (Fig. 6).

Fig. 3.

Early but not late postnatal hyperoxia induces right ventricular hypertrophy (RVH). Fourteen-day-old mice exposed to early PN hyperoxia demonstrate increased Fulton's index [weight of the right ventricle divided by the combined weight of the left ventricle and the septum: RV/(LV+S)] compared with normoxic controls. Mice exposed to late PN hyperoxia failed to demonstrate RVH at 14 days. *P < 0.05.

Fig. 4.

Early and late postnatal hyperoxia do not impact vascular development by vessel count. A: analysis of lung tissue from 14-day-old mice exposed to early and late PN hyperoxia did not show a significant difference in number of small pulmonary arteries per high-powered field. B: representative images at ×10 original magnification.

Fig. 5.

Early and late postnatal hyperoxia do not impact pulmonary vascular remodeling by medial wall thickness. A: hematoxylin and eosin-stained sections of 14-day-old mice exposed to early and late PN hyperoxia failed to demonstrate any change in medial wall thickness. B: representative lung images at ×40 original magnification. AW, airway; PA, pulmonary artery.

Fig. 6.

Early but not late postnatal hyperoxia results in increased muscularization of distal pulmonary arteries. α-Smooth muscle actin-stained sections of 14-day-old mice exposed to early PN hyperoxia demonstrated increased muscularization that was not observed with late PN hyperoxia; n = 4. *P < 0.05 compared with minimal.

Neither early nor late PN O₂ impacted somatic growth, as assessed by body weight at 14 days of age (6.25 ± 0.46, 5.81 ± 0.97, and 5.66 ± 0.62 g, for mice exposed to normoxia, early, and late PN O₂, respectively).

Finally, to determine whether these findings are oxygen concentration dependent, experiments were repeated with PN exposure to 90% O₂. These data demonstrate increased RVH and mean AlvA with both 75 and 90% O₂ compared with normoxia (Fig. 7, A and B). However, late exposure to 90% also trended toward significance and no longer was statistically different from early PN O₂ at 90%.

Fig. 7.

Early and late PN hyperoxia, 75 vs. 90% O₂. A: 14-day-old mice exposed to early PN hyperoxia at 75 and 90% O₂ demonstrate significant increase in RVH compared with normoxic controls. Although Fulton's index in mice exposed to 75% O₂ from P0-3 was greater in than those exposed P4-7, there was no statistical difference with early or late O₂ at 90%. B: 14-day-old mice exposed to early PN hyperoxia at 75 and 90% O₂ demonstrate significant increase AlvA compared with normoxic controls. Although AlvA in mice exposed to 75% O₂ from P0 to P3 was greater than those exposed P4-7, there was no statistical difference with early or late O₂ at 90%; n = 4 for early PN at 90%, n ≥ 6 for other groups. *P < 0.05 compared with normoxia. †P < 0.05 compared with early PN O₂.

MitoTEMPO attenuates compromised alveolarization and RVH during early PN hyperoxia.

To determine whether compromised alveolarization and RVH is mitochondrial ROS dependent, mice exposed to early PN O₂ were treated daily with mitoTEMPO (MT) or PBS vehicle (V) during hyperoxic exposure. Treatment with mitoTEMPO abrogated the effects on alveolarization at 14 days (relative AlvAs of 408.1 ± 23.7, 432.9 ± 14.2, 610.2 ± 36.4, and 466.6 ± 39.2 for normoxia + V, normoxia + MT, hyperoxia + V, and hyperoxia + MT, respectively; Fig. 8). Similarly, mitoTEMPO attenuated the increase in RVH observed with early PN O₂ (Fig. 9).

Fig. 8.

MitoTEMPO protects against hyperoxia-induced alveolar simplification. A: 14-day-old mice treated with mitoTEMPO daily during early PN hyperoxia demonstrate an attenuation of hyperoxia-induced increase in AlvA. *P < 0.05 compared with normoxia + vehicle. †P < 0.05 compared with hyperoxia + vehicle. B: representative lung images at ×20 original magnification.

Fig. 9.

MitoTEMPO protects against hyperoxia-induced RVH. Fourteen-day-old mice treated daily with mitoTEMPO daily during early PN hyperoxia demonstrated attenuation of RVH. *P < 0.05 compared with normoxia + vehicle.

Treatment with mitoTEMPO did not alter mortality or compromise growth. Average weight at 14 days was 6.29 ± 0.85, 6.24 ± 0.62, 6.56 ± 0.32, and 6.66 ± 0.34 g for normoxia + V, normoxia + MT, hyperoxia + V, and hyperoxia + MT, respectively.

Induction of NOX1 and MnSOD during hyperoxia is developmentally regulated and mitochondrial ROS dependent.

We previously demonstrated that the increase in NOX1 expression in response to hyperoxia is developmentally regulated (12). Treatment with mitoTEMPO during early and late PN O₂ was performed in the present study to determine whether age of neonatal exposure or mitochondrial ROS regulated expression of NOX1.

Western blot analysis demonstrated an induction of NOX1 during early PN O₂ that was attenuated by administration of mitoTEMPO with relative expression of 1 ± 0.1, 1.2 ± 0.3, 2.2 ± 0.5, and 0.8 ± 0.2 for normoxia + V, normoxia + MT, hyperoxia + V and hyperoxia + MT, respectively (Fig. 10). By contrast, late PN O₂ failed to induce expression of NOX1 and expression was not affected by mitoTEMPO administration (Fig. 11). In addition, neither exposure to early PN O₂ nor treatment with mitoTEMPO altered expression of NOX2 or NOX4 in response to hyperoxia (data not shown).

Fig. 10.

MitoTEMPO attenuates induction of NOX1 expression during early PN hyperoxia. A: analysis of whole lung protein extract from mice treated with mitoTEMPO or vehicle during early PN hyperoxia demonstrates attenuation of NOX1 induction with mitoTEMPO treatment. Results are expressed as fold change NOX1/β-actin expression vs. vehicle normoxic control. *P < 0.05 compared with normoxia + vehicle. †P < 0.05 compared with hyperoxia + vehicle. B: representative Western blot.

Fig. 11.

Late PN hyperoxia fails to induce NOX1 expression. A: although early PN hyperoxia induces NOX1 expression (Fig. 10), analysis of whole lung extract from mice exposed to late PN hyperoxia showed no significant increase in NOX1 expression or change with mitoTEMPO treatment. Results are expressed as fold change NOX1/β-actin vs. vehicle normoxia control. B: representative Western blot.

As MnSOD plays a key role in the redox status of the mitochondria, we investigated the effect of hyperoxia and mitoTEMPO on MnSOD expression. Similar to findings with NOX1, early PN O₂ caused an induction of MnSOD; however, mitoTEMPO failed to attenuate this increase with relative expression of 1 ± 0.1, 1.3 ± 0.2, 2.4 ± 0.4, and 1.8 ± 0.4 for normoxia + V, normoxia + MT, hyperoxia + V and hyperoxia + MT, respectively (Fig. 12). By contrast, the change in MnSOD expression in late PN hyperoxia failed to reach statistical significance (Fig. 13). Additional analysis of antioxidant regulation revealed that neither exposure to early PN O₂ nor treatment with mitoTEMPO altered the expression of CuZnSOD (data not shown).

Fig. 12.

MitoTEMPO attenuates induction of manganese superoxide dismutase (MnSOD) expression during early PN hyperoxia. Analysis of whole lung protein extract from mice treated with mitoTEMPO or vehicle during early PN hyperoxia demonstrates attenuation of MnSOD induction with mitoTEMPO treatment. Results are expressed as fold change MnSOD/β-actin expression vs. vehicle normoxic control. *P < 0.05 compared with normoxic control. B: representative Western blot.

Fig. 13.

Late PN hyperoxia fails to induce MnSOD expression. A: although early PN hyperoxia induces MnSOD expression (Fig. 12), analysis of whole lung extract from mice exposed to late PN hyperoxia showed no significant increase in MnSOD expression or change with mitoTEMPO treatment. Results are expressed in fold change MnSOD/β-actin vs. vehicle normoxic control. B: representative Western blot.

DISCUSSION

Although a critical window of susceptibility to hyperoxia has previously been postulated for the newborn, the present study in neonatal mice reveals that the effect of supraphysiological oxygen on cardiopulmonary development is determined by age at the time of exposure (5, 22). Specifically, early PN exposure (from P0-3) results in compromised alveolarization, decreased septation, increased distal artery muscularization, and right ventricular hypertrophy at 14 days that is not seen when exposure is delayed by a few days into the PN period (P4-7). Although publications have demonstrated compromised alveolar or vascular development with delayed hyperoxia, these models used higher oxygen concentrations and/or longer duration of exposure (41, 42). We intentionally chose a brief (72 h) exposure to highlight an early window of susceptibility.

Although our findings of right ventricular hypertrophy after early PN hyperoxia were suggestive of pulmonary hypertension, evaluation of the pulmonary vasculature failed to identify compromised vascular development by vessel count or MWT. We recognize the well-described limitations of these techniques in evaluating pulmonary vascular development (10, 47). The small size of our model restricts ability to fix vessels in a perfused state, which may affect assessment of MWT. Vessel count would also ideally be normalized to parenchymal lung volume. Ultimately, documentation of the functional microvascular bed by advanced imaging techniques would provide the most elegant method of documenting the effects on vascular development in our model. However, these advanced imaging techniques are also limited by the small size of the P14 mouse.

It is possible that longer exposure or higher concentration of oxygen would allow detection of altered vascular development after early PN O₂. As an example, Yee et al. (49) exposed rodents to 100% O₂ and observed both RVH and vascular remodeling after 72 h of neonatal exposure. Our results are consistent with those published by Lee et al. (29), who found RVH without changes in pulmonary artery wall thickness after 24 h of 75% O₂. We speculate that the RV remodeling we observed after early PN O₂ could be due to changes in muscularization of small-resistance pulmonary arteries or hyperreactive vasculature. Alternatively, these findings could represent the direct effect of supraphysiological oxygen on the developing heart (25).

We previously demonstrated that both mitochondrial oxidative stress and NOX1 induction in response to hyperoxia are developmentally regulated (12). The NADPH oxidase family plays a role in potentiating oxidative stress and could contribute to the effects of hyperoxia on the developing lung (8). Studies in NOX1-deficient mice have shown that the protein plays a critical role in hyperoxia-induced lung injury in the adult (15). Neonatal models have also implicated NOX enzymes by demonstrating improved type 2 alveolar cell proliferation in p47(phox) knockout mice exposed to hyperoxia (5). Our data reveal that NOX1 induction occurs in response to early but not late PN O₂. This induction with early exposure could then lead to an amplification of ROS signaling that contributes to the observed defects in lung and pulmonary vascular development. Furthermore, we now show that NOX1 induction in hyperoxia is mitochondria ROS dependent, since expression was normalized in mitoTEMPO-treated mice. We suggest that at least some of the lung developmental defects caused by hyperoxia in the immature lung could result from exaggerated mitochondrial oxidative stress response and amplification of ROS signaling via NOX1 activity.

As NOX1 is localized to caveolae on the plasma membrane and endosomes, we speculate that ROS leak from the matrix may play a role in this process (14). In support of that hypothesis, we have previously demonstrated in isolated pulmonary artery smooth muscle cells that hyperoxia increases mitochondrial ROS followed by cytosolic ROS. In those studies, we identified that blocking the mitochondrial matrix ROS with either MitoTEMPO or mitochondria-targeted catalase was sufficient to block the downstream cytosolic effects (21). We see a similar phenomenon here where blocking the mitochondrial matrix ROS is sufficient to block both increased NOX1 expression as well as the phenotypic changes of hyperoxia-induced lung injury. These mechanisms are consistent with those described by Rathore et al. (40) with hypoxia. Those studies demonstrated that inhibition of mitochondrial ROS generation with rotenone or myxothiazol blocked hypoxia-induced increase in NOX activity (40). Translocation of signaling molecules from the mitochondria can also result in activation of intermediary proteins or transcription factors that impact downstream signaling targets. Indeed, Katsuyama et al. (28) demonstrated that NOX1 transcriptional activation is regulated by activating transcription factor-1 (ATF-1), which is triggered by platelet-derived growth factor (PDGF) and prostaglandin F₂α in a mitochondria-dependent manner. Lee et al. (30) further described mitochondrial ROS-dependent NOX1 activation via the PI3K/Rac1 pathway.

In addition to developmental differences in NOX1 regulation, we identified increased MnSOD expression after early but not after late PN hyperoxia. These findings suggest that the immature lung may have mechanisms to respond to the increased burden of oxidative stress, although no significant difference was observed between the mitoTEMPO- and vehicle-treated mice in hyperoxia. However, controversy exists regarding the ability of MnSOD to mitigate lung injury in response to hyperoxia. Studies in mice carrying a transgene to express SOD2 under the control of the β-actin promotor failed to demonstrate prolonged survival with hyperoxia (26). In contrast, studies with transgenic mice carrying a human MnSOD driven by a surfactant protein C promoter demonstrated protection (48).

Collectively, these data indicate that ROS signaling plays a critical role in the disruption of neonatal lung development induced by hyperoxic challenge. These findings suggest that mitochondria-targeted antioxidant therapy might attenuate the consequences of high inspired oxygen levels that may be necessary in treating hypoxemic premature infants who are at risk of developing BPD. Although antioxidants have been tested for a wide range of disease processes, nearly all of the clinical trials of antioxidant therapies have failed to detect any protective effect. One factor contributing to these failures is that broad-based antioxidants can exhibit off-target effects and disrupt normal cell redox-dependent processes (9, 18, 44). In that regard, low levels of ROS are involved in signal transduction systems that regulate cell proliferation, protective responses, and redox homeostasis, as well as innate immunity and apoptosis (14, 31). Adverse effects of antioxidants have been identified in animal models with evidence of increased lung injury and pulmonary morbidity (13, 37, 43). Few studies using antioxidants have been performed in premature infants because of the concerns for potential risks. Some promising data were reported by Davis et al. (20), who tested intratracheal administration of recombinant human CuZnSOD as a preventative measure for BPD. Although that study failed to demonstrate protection against BPD at 36 wk, long-term follow-up revealed a decreased incidence of pulmonary morbidities and reduced rates of rehospitalization at 1 year (19, 33, 38).

Because mitochondria are key modulators of oxidative stress in response to hyperoxia, we suggest that a mitochondria-targeted antioxidant may protect against lung injury and compromised cardiopulmonary development seen in BPD while minimizing potential off-target effects. Our data demonstrate that treatment with the mitochondria-targeted antioxidant, mitoTEMPO, abrogated in full the negative effects of hyperoxia on both lung and vascular development. Our evaluations of lung development were carried out at 14 days because the size of the mice did not permit reliable assessments of lung morphology before that time. Our findings at 14 days could therefore reflect a difference in the rate of recovery and could underestimate the immediate effects on lung development. However, because heart and lung tissue were harvested at 14 days for both groups, mice exposed later had a shorter recovery time yet were less affected by the hyperoxic exposure. These mice still demonstrated improved alveolarization and absence of right ventricular remodeling, which suggests that our findings were related to the direct effects on cardiopulmonary development rather than on recovery. Although we did not observe any increase in mortality or compromised growth with this treatment, alternative routes of delivery or tissue-specific targeting might further limit the off-target effects on nonpulmonary tissue. Intratracheal delivery should be considered with studies implicating the alveolar epithelial compartment as a major site of antioxidant protection in the neonatal period (39).

In summary, these studies are the first to show that the effects of hyperoxia on the immature lung are developmentally regulated and mitochondrial ROS dependent. Our findings reaffirm the need for judicious use of oxygen therapy in premature infants at risk of developing BPD. NOX1 induction with early exposure suggests that amplification of ROS signaling may contribute to the susceptibility observed with early as opposed to late PN exposure. Finally, our data from mitoTEMPO experiments argue the critical need for further investigation of targeted antioxidant therapy for both prevention and treatment of BPD.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grants HL35440 and HL 122062 (P. T. Schumacker) and HL109478 (K. N. Farrow).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

A.D., S.F.G., and S.K.B. conception and design of research; A.D., G.A.K., J.M.T., S.F.G., and S.K.B. performed experiments; A.D., G.A.K., J.M.T., S.F.G., P.T.S., and S.K.B. analyzed data; A.D., G.A.K., S.F.G., K.N.F., P.T.S., and S.K.B. interpreted results of experiments; A.D. and S.K.B. prepared figures; A.D., K.N.F., P.T.S., and S.K.B. drafted manuscript; A.D., G.A.K., K.N.F., P.T.S., and S.K.B. edited and revised manuscript; A.D., G.A.K., J.M.T., S.F.G., K.N.F., P.T.S., and S.K.B. approved final version of manuscript.