Acute and chronic changes in the control of breathing in a rat model of bronchopulmonary dysplasia

Abstract

Infants born very prematurely (<28 wk gestation) have immature lungs and often require supplemental oxygen. However, long-term hyperoxia exposure can arrest lung development, leading to bronchopulmonary dysplasia (BPD), which increases acute and long-term respiratory morbidity and mortality. The neural mechanisms controlling breathing are highly plastic during development. Whether the ventilatory control system adapts to pulmonary disease associated with hyperoxia exposure in infancy remains unclear. Here, we assessed potential age-dependent adaptations in the control of breathing in an established rat model of BPD associated with hyperoxia. Hyperoxia exposure (${\text{FI}}_{{\text{O}}_{\text{2}}}$; 0.9 from 0 to 10 days of life) led to a BPD-like lung phenotype, including sustained reductions in alveolar surface area and counts, and modest increases in airway resistance. Hyperoxia exposure also led to chronic increases in room air and acute hypoxic minute ventilation (V̇e) and age-dependent changes in breath-to-breath variability. Hyperoxia-exposed rats had normal oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$) in room air but greater reductions in $\text{S}{\text{p}}_{{\text{O}}_2}$ during acute hypoxia (12% O₂) that were likely due to lung injury. Moreover, acute ventilatory sensitivity was reduced at P12 to P14. Perinatal hyperoxia led to greater glial fibrillary acidic protein expression and an increase in neuron counts within six of eight or one of eight key brainstem regions, respectively, controlling breathing, suggesting astrocytic expansion. In conclusion, perinatal hyperoxia in rats induced a BPD-like phenotype and age-dependent adaptations in V̇e that may be mediated through changes to the neural architecture of the ventilatory control system. Our results suggest chronically altered ventilatory control in BPD.

INTRODUCTION

Premature infants often have respiratory distress at birth due to pulmonary immaturity. They typically require interventions such as supplemental oxygen therapy (hyperoxia) to improve their oxygen saturations ($\text{S}{\text{p}}_{{\text{O}}_2}$). Although use of oxygen is critical for survival, a negative consequence of long-term exposure to perinatal hyperoxia is the development of bronchopulmonary dysplasia (BPD), a chronic lung disease affecting ∼10,000 infants/yr in the United States (36). Histologically, BPD is characterized by reduced alveolar surface area and pulmonary vascular development, effectively impairing gas exchange (27). As a result, infants with BPD often require additional supportive respiratory therapies, leading to increased risk of respiratory complications and respiratory-related mortality.

Premature infants at risk for BPD also suffer from respiratory control abnormalities, indicated by periodic breathing (43), apnea of prematurity (17), and reduced ventilatory responses to hypoxia (2, 10). BPD infants have significantly different breathing frequencies and tidal volumes compared with age-matched non-BPD infants (16). BPD infants that survive after hospital discharge have a greater risk of dying from sudden infant death syndrome (SIDS), which is the unexpected death of an infant < 12 mo of age, likely due to impaired cardiorespiratory control (54). Adults with BPD are reported to have decreased exercise tolerance, which may be due to both impaired lung growth and/or neural control of breathing (8, 46). Despite the apparent acute and chronic changes to control of breathing in infants with BPD, the age-related adaptations in ventilatory control and potential changes within the brainstem nuclei controlling breathing have not been well studied.

Previous investigations of the impact of neonatal hyperoxia exposure on the control of breathing resulted in largely equivocal results as it pertains to eupneic ventilation. Ling et al. (30, 31) and others (3, 4, 6, 14) demonstrated that neonatal hyperoxia exposure (but not adult hyperoxia) blunted the development of carotid bodies, the primary chemoreceptors sensing hypoxia. This may consequently blunt the hypoxic ventilatory response, providing a mechanism to explain the blunted hypoxic ventilatory sensitivity in infants with BPD (10, 22, 29). However, other studies have reported that neonatal hyperoxia (60% inspired O₂) did not alter eupneic ventilation (53), whereas some have reported a resting hyperventilation following chronic hyperoxia (30). Finally, shorter exposures to perinatal hyperoxia (E18 to P5) led to decreases in phrenic nerve discharge (7). Thus, the resultant effects of chronic perinatal hyperoxia on the control of breathing have not been well established. Furthermore, many of these studies have not characterized the extent of pulmonary dysfunction that accompanies these ventilatory changes following chronic perinatal hyperoxia.

As suggested by rodent studies, hyperoxia exposure also negatively impacts brain development through effects on neurons, oligodendrocytes, astrocytes, and microglia (45). Levels of hyperoxia shown to result in BPD in premature infants have been associated with hypomyelination and apoptosis leading to white matter microstructural changes in the brain (20). These and other potential central nervous system alterations that result from hyperoxia exposure may functionally manifest in a variety of ways, including hyperactivity, cognitive impairments (24, 48), and encephalopathy of prematurity (11). Whether similar brain changes occur within the respiratory control network after hyperoxia-induced BPD remains unknown.

In the current study, we tested the hypothesis that the control of breathing is acutely and/or chronically altered following chronic perinatal hyperoxia exposure in rats using an established protocol (26, 42, 52) by measuring breathing and indices of oxygenation ($\text{S}{\text{p}}_{{\text{O}}_2}$ and/or blood gases) at rest and during acute hypoxia. Furthermore, we sought to determine whether there are long-term anatomic changes in neurons or astrocytes within several key brainstem nuclei controlling breathing that occur directly or indirectly because of chronic perinatal hyperoxia exposure.

METHODS

Animals

The Animal Care and Use Committee at the Medical College of Wisconsin (MCW) approved all experimental protocols before the initiation of experiments. Timed-pregnant Sprague-Dawley dams were purchased from a commercial supplier (E15–16 on arrival; Envigo, Madison, WI) and were housed in the Biomedical Resource Center at MCW until the birth of the litter, at which time the dam and pups were placed in the chamber for exposure to hyperoxia (${\text{FI}}_{{\text{O}}_{\text{2}}}$ >90% for 10 days) or normoxia (room air; 10 days). A total of 80 rat pups were used (35 normoxic and 45 hyperoxic rats) from five control (normoxic) and six hyperoxic litters. Not all rats per litter were used, and the average litter was kept similar between normoxic and hyperoxic rats (10–12 rats/litter). Rats were maintained on a 12-h:12-h light-dark cycle, and food and water were given ad libitum.

Hyperoxia and Normoxia Exposure

On the day of birth, rat litters and their dams were moved into a chamber supplied with 90% oxygen from compressed oxygen tanks with a continuously running fan and small leak to prevent CO₂ buildup. The chamber was opened twice daily for <5 min to remove the rat dam for 2 h each day to maintain overall maternal health. After 10 days of hyperoxia exposure, physiological and tissue studies began at postnatal day (P)10. Thereafter, rats were maintained under room air conditions for the remainder of the study. Normoxic rats were always in room air conditions until completion of the study.

Histology

Tissue collection.

Rats were anesthetized with isoflurane in propylene glycol (20% vol/vol). Lungs were flushed through the right ventricle with 20–50 ml of PBS. Then, the brain was flushed through the left ventricle with 200 ml of PBS. The trachea was isolated, and lungs were inflation fixed in situ at 25 cmH₂0 pressure with 10% formalin for 25 min. During that time, the brain was fixed through the left ventricle with 200 ml of 4% paraformaldehyde. Tissues were removed, and lungs were placed in 10% formalin overnight and then dehydrated in 70% EtOH until they were needed for lung volume estimation by the Cavalieri method (9, 25) before paraffin embedding, sectioned at 4 µm, and mounted onto SuperFrost Plus slide (Denville Scientific, Metuchen, NJ). Brain tissue was postfixed for 24 h in 4% PFA in PBS and then dehydrated with graded sucrose (10, 20, and 30% sucrose) before being frozen (−80°C), coronally sectioned (20 µm), and mounted to SuperFrost Plus slides.

Lung histology and measurements.

After dehydration and before paraffin embedding, starting with a random orientation, lungs were fully sectioned at 5 mm under room air conditions for lung volume estimation by the Cavalieri method (25). In Fiji (ImageJ) (47), A 5 × 5 mm grid was overlaid onto an image of all lung sections made for each animal, and the number of intersections falling on tissue was counted. The number of intersections times the area of each grid (25 mm²) times the section thickness (5 mm) was used to calculate lung volume. Every third lung section (the 1st section randomly selected) was chosen for paraffin embedding into a single block, effectively employing systematic uniform random sampling (SURS). Two sets of serial lung sections were made (4 total lung sections/rat), with ≥200 µm of distance between each set. Lung sections were deparaffinized and then stained with hematoxylin and eosin. Using an E80i Nikon light microscope and ×10 objective, two regions per lung section were randomly selected (i.e., SURS-based) for imaging and considered “reference” images. The same region on the corresponding serial section was imaged and considered to be the “lookup” image, allowing for the physical dissector methodology to be employed to quantify number of alveoli (9, 25). All images were analyzed using Fiji. With a point grid (area/point: 10,000 µm²) that was randomly offset, the number of points falling on parenchyma and nonparenchymal portions of each reference image were counted to assess the parenchymal volume (total no. of parenchymal points/total points of grid × lung volume determined by Cavaleiri method) as outlined by others (9). Alveolar volume and surface area were next quantified using a test line system (length: 150 µm; length/point: 75 µm) superimposed onto all reference images taken; all septal intersections within each test line were counted as well as the total number of line end points that fell on alveolar lumen or septa. Alveolar airspace volume and total septal surface area (i.e., gas exchange area) were calculated according to Brandenberger et al. (9). Finally, the physical dissector methodology was employed to estimate the number of alveoli as per the American Thoracic Society and described by others (9, 25). All pairs (n = 8/animal) of serial images were loaded into the TrakEM2 plugin for Fiji, the two images were aligned, and a 400 × 400 µm square was superimposed directly in the middle of each image. The number of alveoli forming (or “bridges” appearing) when toggling from the reference to the lookup image within the square (and if falling along the top and right sides) were counted and used to estimate the number of alveoli.

Immunofluorescence and immunohistochemistry.

Brainstem tissue sections were stained using minor modifications to an established staining protocol (44). Briefly, tissue sections were placed into a drying oven (57°C) for 2 h and then treated with antigen retrieval (citrate buffer) for 4 h at 70°C, washed with PBS (3 × 10 min), and then blocked (5% normal horse serum) for 1 h. Brain tissue sections were placed in 2.5% serum with GFAP primary antibody (1: 1,000; rabbit polyclonal, AB1234; Sigma) or NeuN primary antibody (1:200,mouse monoclonal, MAB3777; Millipore) overnight at 4°C. Sections were washed (3 × 10 min PBS). GFAP-stained tissue was then incubated in Dylight 594-conjugated to anti-rabbit secondary antibody (1:500; 45 min; Vector), washed (3 × 10 min PBS), and then coverslipped. NeuN-stained tissue was incubated for 1 h in horse anti-mouse biotinylated secondary antibody (1:1,000, BA2000; Vector), washed (3 × 10 min PBS), incubated in ABC reagents (30 min; PK-6100, Vector) rinsed (3 × 5 min PBS), exposed to DAB (45 s, K3468; DAKO North America), rinsed, dehydrated with graded ethanol (70, 70, 95, and 100% for 1 min each) before being cleared in Xylenes (2 min), and then coverslipped.

GFAP and NeuN analyses.

Every 400 µm, a representative image of the nuclei raphe obscurus (ROB), raphe magnus (RMG), nucleus of tractus solitarus (NTS), nucleus ambiguus (NA), retrotrapezoid nucleus (RTN), and the ventral respiratory column (VRC) was captured for GFAP and NeuN analyses guided by the Rat Brain Atlas (Paxinos and Watson, 5th ed.), except that the retrotrapezoid nucleus (RTN) was identified based on the findings of Stornetta et al. (51). Images of immunolabeled sections were captured using an E400 Nikon epifluorescence microscope with a ×20 objective using Metavue research imaging system software (Molecular Devices). Exposure time was kept constant for all images. The excitation and emission filters for the labeled dyes were as follows: DAPI (350; 440/25 nm) and Dylight-594 (560; 600/25 nm). Images were opened in Fiji and converted to an 8-bit format, a macro was ubiquitously applied to enhance contrast and decrease background, and then an auto-local thresholding was completed using the Fiji-based Phansalker method. The image was converted to a binary image, and the ratio of positive signal relative to the total number of pixels in the image was calculated. All GFAP measurements from hyperoxia tissue sections were normalized to co-stained normoxia tissue sections to be expressed as a percent of control.

Images of DAB-stained sections were imaged using an E80i Nikon light microscope via a ×20 objective. Images were opened in Fiji (National Institutes of Health, Bethesda, MD) software, a standardized region was used to analyze a specific nucleus, the image was converted to an 8-bit image, and an auto-local threshold was applied (using Phansalker method in Fiji) and then converted to binary. A watershed was applied, and the number of objects fitting a certain size (100–3500 μm² and 0.1 to 100 in circularity) were counted to represent the number of neurons. All neuron counts from hyperoxia tissue sections were normalized to co-stained normoxia tissue sections and are expressed as a percent of control.

Physiological Measurements

Two sets of plethysmographs were used. One set included commercially available Buxco neonatal and adult whole body plethysmographs using FinePointe software to collect and analyze the data using proprietary and built-in algorithms to measure indices of airway resistance and timing of respiratory phases (Data Sciences International, St. Paul, MN). These plethysmograph systems have pneumotach screens that allow air to flow through but still cause a measurable pressure deflection within the chamber relative to outside of the chamber. Thus, flow is a measured parameter during inspiration and expiration. Rats 10–11, 20–21, and 60 days old were exposed to 12 min of room air, and the last 5 min were used for analysis. A range of 2 days were used for these measurements because it was not feasible to study all animals on a single day. If rats were not calm after the first 7 min of the study, then an additional 7 min was added.

The second set of plethysmographs were custom-built and used in prior studies to measure neonatal (28) and adult (37) ventilation, permitting more control over chamber temperature, flow rate, air type (room air vs hypoxia), and data analysis (i.e., easier customizable calculations) via LabChartPro recording and analysis software. Chamber temperature and air flow were continuously measured. Atmospheric pressure, body weight, and temperature were measured before and after the study. Ventilatory measurements in rat pups (P10–P21) were made using a custom-built, 200-ml plethysmograph, as described previously (28). Briefly, gas inflow and outflow rate were maintained at ∼150 ml/min. The aluminum floor was temperature controlled (29–30°C; Dyna-sense; Scientific Instruments) to maintain body temperatures in rat pups, as described previously (28). Chamber O₂ and CO₂ (O₂ capnograph, 07-0193), chamber pressure (Validyne differential pressure transducer), and temperature (∼27–30°C; Warner Instruments) were continuously measured. Relative humidity was assumed to be constant at 20% based on prior measurements (28). Analog signals were converted to digital signals and recorded using Windaq data acquisition software (200-Hz sampling rate). Windaq files were opened in Laboratory chart software for analysis. Rectal temperatures were measured before and after each study using a T-type thermocouple probe and reader (Omega).

Ventilatory measurements in adult rats (P43 and P60) were made using a custom-built, 10-liter plethysmograph, as described previously (28, 37, 38). Briefly, gas inflow rate was maintained at 10 l/min, whereas the outflow was slightly greater to avoid CO₂ buildup. The chamber floor was not heated. Chamber O₂ and CO₂ (O₂ capnograph; 07-0193), chamber pressure (Validyne differential pressure transducer), temperature (Warner Instruments), and relative humidity (HX93-A; Omega) were continuously measured. All analog signals were converted to digital signals and recorded using Laboratory chart data acquisition software (1-kHz sampling rate). Laboratory chart software was used for analysis. Rectal temperatures were measured before and after each study using a J-type thermocouple probe and reader (BAT-12; Life Science Instruments).

Breathing of neonatal (P10, P12, P14, P17, and P21) and adult (P43 and P60) rats was recorded while rats were exposed to 20 min of room air (21% O₂, balanced N₂), followed by 10 min of acute hypoxia (12% O₂, balanced N₂). The last 10 min of room air and the last 5 min of acute hypoxia were used for ventilatory analyses (see below). These portions of the recording were analyzed because the rats were most stable during either condition and because switching to hypoxia gas requires <5 min to stabilize to 12% within the chambers.

$\text{S}{\text{p}}_{{\text{O}}_2}$ Measurements.

A MouseOx (MouseOx Plus; STARR Life Sciences) with the appropriately sized collar sensor was used to measure $\text{S}{\text{p}}_{{\text{O}}_2}$ in conscious P12, P14, P17, P21, and P43 rat pups. The collar was tunneled through our plethysmographs before each study and then placed around the neck of the rat. An adequate $\text{S}{\text{p}}_{{\text{O}}_2}$ signal was verified before each study. The fur around the neck of P43 rats was shaved for an adequate $\text{S}{\text{p}}_{{\text{O}}_2}$ signal. $\text{S}{\text{p}}_{{\text{O}}_2}$ data was collected via Windaq software (DATAQ Instruments, Akron, OH) during ventilation recording. Data were processed and analyzed offline in Microsoft Excel.

Blood gas measurements.

Adult P50 rats underwent femoral artery catheterization, as previously described (37). In brief, using aseptic techniques, rats were (39) anesthetized with isoflurane, and upon loss of the righting reflex, rats were placed on a heated pad with a nose cone to maintain anesthesia (2.5% isoflurane). The femoral artery was isolated and lifted, and a sterile, prefabricated femoral catheter was tunneled ∼3 cm into the animal. The other end of the catheter was tunneled subcutaneously and externalized between the scapulae, where it was anchored to the muscle. Three centimeters of tubing, backfilled with heparinized saline, was allowed to remain externalized. Proper intraoperative and postoperative medications were given. At P60, the steel stopper preventing blood outflow from the catheter was removed, and additional tubbing was connected from steel tubing adapters to the externalized catheter and a blunt needle traversing the wall of the plethysmograph. Arterial blood samples (0.4 ml) were collected in a heparin-coated 1-ml syringe between the last 3 and 5 min of room air and hypoxia conditions. Blood gases were measured with a Rapid Laboratory Model 248 blood gas analyzer (Bayer Healthcare, Leverkusen, Germany).

Data Analyses

All data collected using the Buxco system were analyzed using FinePointe Software (DSI) built-in analyses tools for drive to breathe [tidal volume (V_T; ml/breath)/inspiratory time (T_I; s)], 50% of expiratory flow (EF50), end inspiratory and expiratory pause (ms), peak inspiratory and expiratory flow (ml/s), and inspiratory and expiratory time (s). All data were analyzed using LabChart. Briefly, two additional channels were added to each file. The first channel used the “cyclic measurements” option to measure the voltage deflection from peak to valley of each respiratory cycle peak, which was subsequently converted from volts to milliliters using a volume calibration and corrected for animal and chamber temperature, relative humidity, and ambient barometric pressure to estimate V_T per breath (15, 23). The second channel used the “cyclic measurements” option to measure the rate of each respiratory cycle (i.e., breathing frequency). These calculations were made from multiple periods of raw data with no less than 200 cumulative breaths per condition. The selected data were void of apneas, sighs, movement, and sniffing artifacts. Minute ventilation (V̇e; ml/min) was calculated from the product of breathing frequency (f; breaths/min) and tidal volume (V_T; ml/breath). For minute ventilation, tidal volume, and breathing frequency no sex differences were measured, and therefore, all data sets presented represent pooled male and female groups.

Poincare analyses were used to assess ventilatory variability for tidal volume and interbreath interval (IBI), which is a function of breathing frequency. Poincare analyses were completed using equations outlined by others (35). The distance of the short radius (SD1) is a measurement of short-term variability or the variability on a breath by breath basis. The distance of the long radius (SD2) is a measurement of long-term variability or the variability of breaths within the entire collection of breaths. For either IBI or V_T Poincare plots, the following equations were used to calculate SD1 and SD2 per rat:

$$X1=\frac{\left(breathn\right)-\left(breathn+1\right)}{\sqrt{2}}$$

$$\text{SD}1=\sqrt{samplevarianceofallX1values}$$

$$X2=\frac{\left(breathn\right)+\left(breathn+1\right)}{\sqrt{2}}$$

$$\text{SD}2=\sqrt{samplevarianceofallX2values}.$$

The breath variable represents an IBI or V_T value. A single SD1 and SD2 value was calculated for each animal during room air and hypoxia breathing.

Statistics

Statistics were run using a variety of software programs. Prism 7 (GraphPad) software was used to quantify lung and brain histology and P60 blood gases by unpaired t-tests. SAS version 9.4 (SAS, Cary, NC) was used to run a mixed model adjusted for repeated measures on all physiological data (ventilation, MouseOx, Poincare analyses), because some but not all animals were studied at each age, and therefore, traditional repeated-measures statistics could not be employed. Bonferroni post hoc analyses were performed as warranted. Significance thresholds were set to P < 0.05. Data are expressed as means ± SE.

RESULTS

Body weights (g) were measured and compared within each age studied (unpaired t-test) and were significantly lower (P < 0.05) in hyperoxic rats compared with normoxic rats at P10 (13.4 ± 3.7 vs. 18.0 ± 4.7 g), P12 (16.5 ± 3.7 vs. 20.6 ± 4.1 g), P14 (20.3 ± 3.6 vs. 24.1 ± 3.8 g), P17 (26.0 ± 3.7 vs. 28.3 ± 2.3 g), and P21 (36.3 ± 4.2 vs. 39.3 ± 3.1 g) but not at P43 (157.4 ± 4.6 vs. 159.4 ± 2.026 g) or P60 (235.2 ± 5.2 vs. 246.0 ± 10.8 g). Four of the 45 rat pups exposed to hyperoxia died between P3 and P7, whereas 0 of the 35 normoxic rats died.

Hyperoxia Exposure Impairs Pulmonary Morphology and Ventilatory Mechanics

Neonatal hyperoxia exposure impaired lung development when qualitatively assessed at P10, P21, and P60 (Fig. 1A). Total lung volume estimates were lower at P10 between hyperoxic and normoxic rats (P < 0.05), but with increasing age there were no differences between the two groups at P21 and P60 (P > 0.05; Fig. 1B). At each of the three ages, hyperoxia-exposed rats had lower septal surface area, alveolar density, and number of alveoli (P < 0.05; Fig. 1, C–E). These changes indicate a sustained effect of neonatal hyperoxia in arresting lung development, specifically reduced alveolarization, and pulmonary simplification.

Fig. 1.

Perinatal hyperoxia exposure induced a bronchopulmonary dysplasia (BPD)-like phenotype assessed by total lung volume, septal surface area, alveolar density, and alveolar counts. A: representative ×10 hematoxylin and eosin images of lung sections from normoxic and hyperoxic rats at postnatal day (P)10, P21, and P60. B–E: total lung volume was significantly lower only at P10 and not at P21 or P60 (B), whereas septal surface area (C), alveolar density (D), and alveolar counts (E) were significantly lower at P10, P21, and P60 in hyperoxic (n = 7) rats compared with age-matched normoxic (n = 6) rats. *P < 0.05 vs. age-matched normoxic group. Scale bar, 100 µm.

To begin to investigate the physiologic consequences of hyperoxia-induced BPD, including reductions in alveolarization and pulmonary simplification, air-flow-based indices of airway resistance and respiratory phase timing were measured at three ages (P10–11, P20–21, and P60). At P20–21 and P60, peak-inspiratory flow was lower in hyperoxia-exposed rats compared with age-matched normoxic rats (Fig. 2A; P < 0.05). There were no differences between groups at any ages for peak expiratory flow, 50% of expiratory flow (EF50), inspiratory time, and expiratory time (Figs. 2B–E, respectively; P > 0.05). At P20–21 and P60 there was a decrease in Rpef (the ratio of the time to peak expiratory flow divided by the expiratory time; a decreased Rpef suggests an increase in airflow resistance) in hyperoxia-exposed rats compared with age-matched normoxic rats suggestive of an increase in expiratory resistance (Fig. 2F; P < 0.05).

Fig. 2.

Airflow-based plethysmographic metrics indicate a modest increase in airway resistance and no change in respiratory phase timing. A and B: peak inspiratory (A) but not expiratory flow (B) were significantly decreased at postnatal day (P)20–P21 and P60. C–E: the expiratory flow rate at 50% (EF50; C), inspiratory time (D), and expiratory time (E) were unchanged at all ages. F: Rpef (ratio of time to peak expiratory flow divided by expiratory time) was significantly lower in hyperoxic rats at P20–P21 and P60 compared with normoxic rats. P60 values within each group were significantly different from P10 to P11 values for all metrics measured. *P < 0.05 vs. age-matched normoxic group; ^P < 0.05 vs. within-group P10–P11 value. P10–P11: normoxic, n = 15; hyperoxic, n = 7. P20–P21: normoxic, n = 20; hyperoxic, n = 14; P60: normoxic, n = 12; hyperoxic, n = 12.

We next compared the effect of hyperoxia exposure on the developmental pattern of breathing by comparing P10–P11 values with subsequent ages within each treatment group. For all metrics, P60 values were greater than P10 values for both normoxic and hyperoxic rat groups (P < 0.05; Fig. 2, A–F). There were two metrics that had different P20–P21 values compared with P10 values. At P20–P21, both normoxic and hyperoxic rat groups had higher peak inspiratory flow compared with respective P10 values (P < 0.05; Fig. 2A). Finally, peak expiratory flow was greater at P20–P21 compared with P10 values for normoxic but not hyperoxic rats (P < 0.05; Fig. 2B). These data indicate modest increases in airway resistance with hyperoxia exposure.

Hyperoxia Exposure Alters Ventilation at Rest and During Acute Hypoxia Exposure

Room air (eupneic) and acute hypoxic ventilation was altered during development and in adults after perinatal hyperoxia exposure (hyperoxic rats; Fig. 3). An age range that spans an established critical window of respiratory development in rat pups at P12 and P14 (19, 32, 34) and minute ventilation (V̇e; ml/min/100g) during room air and acute hypoxia at P60 were greater in hyperoxia-exposed rats compared with controls (P < 0.05; Fig. 3, A and B). The increases in V̇e for hyperoxic rats during room air was due to a greater breathing frequency and tidal volume (V_T; ml·breath⁻¹·100 g⁻¹) at P12 (P < 0.05) but due only to a greater breathing frequency at P14 (P < 0.05; Fig. 3, C and D). The increases in V̇e for hyperoxic rats during acute hypoxia were due to a greater breathing frequency at P12 (P < 0.05) but due only to a greater breathing frequency and V_T at P14 (P < 0.05; Fig. 3, D and F). The increase in V̇e at P60 during room air and acute hypoxia was due to a greater V_T compared with normoxic rats (P < 0.05; Fig. 3, E and F). V̇e under room air conditions continued to be higher for hyperoxic rats compared with control rats at P17 and P21 (P < 0.05; Fig. 3A). The hypoxic ventilatory response (HVR; hypoxia ventilation as a percent of room air ventilation) was measured and compared within each age studied and was not significantly (P > 0.05) different (except at P21) in hyperoxic rats compared with normoxic rats at P10 (120.6 ± 5.3 vs. 118.3 ± 2.9), P12 (117.2 ± 6.1 vs. 124.4 ± 5.0), P14 (134.5 ± 5.4 vs. 138.8 ± 5.0), P17 (138.7 ± 5.6 vs. 146.8 ± 5.6), P43 (122.2 ± 5.6 vs. 131.3 ± 5.6), or P60 (132.8 ± 6.6 vs. 139.0 ± 10.5). The HVR was different at P21 between the two groups (141.7 ± 4.7 vs. 183.4 ± 8.6, P < 0.05).

Fig. 3.

Neonatal hyperoxia exposure caused acute and chronic changes to ventilation measured during room air and acute hypoxia. There were many differences measured between hyperoxic and normoxic rats at each age during [postnatal day (P)10 to P43] and after (P60) development for minute ventilation (V̇e; A and B), breathing frequency (C and D), and tidal volume (V_T; E and F) during room air and acute hypoxia challenges. Notably, there were consistently higher minute ventilations measured acutely at P12–P14 due to a combination of increased breathing frequency and V_T and chronically at P60 due to an increase in V_T in hyperoxic rats compared with normoxic rats during room air and acute hypoxia. Additionally, developmental patterns in breathing were assessed by comparing measured P10 values with subsequent ages within a group, resulting in a different set of significantly different comparisons within normoxic vs. hyperoxic groups, indicating shifts in the developmental pattern of breathing (A–F). *P < 0.05 vs. age-matched normoxic group; ^P < 0.05 vs. within-group P10 value.

Breathing variables (V̇e, frequency, and V_T) measured at all ages studied were compared with P10 values for normoxic and hyperoxic rat groups to assess the impact of hyperoxia-induced BPD on the developmental pattern of breathing. Overall, hyperoxia-induced BPD acutely and chronically altered the developmental pattern of breathing. V̇e was higher at P12 and P14 compared with P10 values during room air (P < 0.05) and higher at P12–P14 and P60 compared with P10 values during acute hypoxia (P < 0.05). Whereas normoxic rats had lower V̇e during room air and acute hypoxia at P43 and P60 (P < 0.05), hyperoxic rats sustained P10 V̇e levels (P > 0.05; Fig. 3, A and B). Breathing frequency was also higher at P12 and P14 compared with P10 values during room air and acute hypoxia in hyperoxic but not normoxic rats (P < 0.05). There was also a sustained increase in breathing frequency at P60 compared with P10 values for room air breathing in hyperoxic but not normoxic rats (Fig. 3, B and C). Finally, V_T was higher at P12 and P14 compared with P10 values during room air in hyperoxic but not normoxic rats (P < 0.05; Fig. 3D). There were no changes in V_T during acute hypoxia compared with P10 values for either hyperoxic or normoxic rats (P > 0.05; Fig. 3F). Together, these data indicate that ventilatory patterns were significantly altered in an age-dependent manner with perinatal hyperoxia exposure.

Ventilation and Blood Oxygenation After Perinatal Hyperoxia Exposure

To assess whether blood gas homeostasis was altered after hyperoxia exposure, blood oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$; pulse oximeter) or arterial blood gases were measured. Throughout development (P12–P43), $\text{S}{\text{p}}_{{\text{O}}_2}$ was comparable between hyperoxic and normoxic rats while they breathed room air (P > 0.05; Fig. 4A). However, during acute hypoxia, $\text{S}{\text{p}}_{{\text{O}}_2}$ was significantly lower at P12 and P17 in hyperoxia-exposed rats compared with controls (P < 0.05; Fig. 4B). As adults, concurrent $\text{S}{\text{p}}_{{\text{O}}_2}$ measurements and arterial blood gas samples indicated that $\text{S}{\text{p}}_{{\text{O}}_2}$, $\text{P}{\text{a}}_{{\text{O}}_2}$, $\text{P}{\text{a}}_{{\text{C}\text{O}}_2}$, and arterial pH were not different between hyperoxic and normoxic rats during room air breathing or during acute hypoxia (P > 0.05; Fig. 5, A–D). Given minimal HVR changes between normoxic and hyperoxic rats but a difference in $\text{S}{\text{p}}_{{\text{O}}_2}$, we next measured the relationship between ventilation and $\text{S}{\text{p}}_{{\text{O}}_2}$ during room air and acute hypoxia to assess how much ventilation is needed to sustain a given $\text{S}{\text{p}}_{{\text{O}}_2}$. Hyperoxia-exposed rats had significantly greater V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$ ratios at P12, P14, P17, P21, and P43 in room air and significantly greater V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$ ratios at P12, P14, and P17 during acute hypoxia compared with normoxic rats (P < 0.05; Fig. 4, C and D). P60 hyperoxic rats had a greater V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$ ratio during room air and acute hypoxia compared with normoxic rats (P < 0.05; Fig. 5E). The room air V̇e/$\text{P}{\text{a}}_{{\text{O}}_2}$ ratio in P60 rats was not different between hyperoxic and normoxic rats (P > 0.05; Fig. 5F). Together, these data indicate that a greater level of ventilation is needed to maintain a normal $\text{S}{\text{p}}_{{\text{O}}_2}$ in hyperoxia-exposed rats that are likely to overcome the lung disease but that this compensation is insufficient during acute hypoxic challenges. These findings are consistent with impaired pulmonary diffusion and/or V/Q mismatch.

Fig. 4.

Blood oxygen saturation and minute ventilation (V̇e)/ oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$) during development [postnatal day (P)12 to P43] during room air and acute hypoxia. A and B: $\text{S}{\text{p}}_{{\text{O}}_2}$was similar between normoxic and hyperoxic rats during room air breathing (A), but hyperoxic rats had significantly lower $\text{S}{\text{p}}_{{\text{O}}_2}$ from P12 to P17 of age during acute hypoxia (B). C and D: hyperoxic rats had a greater minute ventilation (V̇e) for a given $\text{S}{\text{p}}_{{\text{O}}_2}$ (V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$) from P12 to P43 during room air breathing (C) and from P12 to P17 during acute hypoxia (D). E: hyperoxic rats had lower hypoxic sensitivity (absolute value of the change in ventilation divided by the change in $\text{S}{\text{p}}_{{\text{O}}_2}$ measured during acute hypoxia and room air) at P12–P14. *P < 0.05 vs. age-matched normoxic group; ^P < 0.05 vs. within group P10 value. Normoxic P12, P14, P17, P21, and P43, n = 7, 9, 14, 12, and 12, respectively. Hyperoxic P12, P14, P17, P21, and P43, n = 10, 6, 11, 10, and 14, respectively.

Fig. 5.

Blood oxygen, blood gas, and minute ventilation (V̇e)/oxygen saturation ($\text{S}{\text{p}}_{{\text{O}}_2}$) and ventilation/$\text{P}{\text{a}}_{{\text{O}}_2}$ at postnatal day (P)60 during room air (RA) and acute hypoxia. A–D: $\text{S}{\text{p}}_{{\text{O}}_2}$, $\text{P}{\text{a}}_{{\text{O}}_2}$, $\text{P}{\text{a}}_{{\text{C}\text{O}}_2}$, and pH were not different between normoxic and hyperoxic rats for either room air or hypoxia breathing. E and F: hyperoxic rats had a greater V̇e for a given $\text{S}{\text{p}}_{{\text{O}}_2}$ (V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$) during room air and acute hypoxia breathing (E), and V̇e relative to a given $\text{P}{\text{a}}_{{\text{O}}_2}$ (V̇e/$\text{P}{\text{a}}_{{\text{O}}_2}$) was greater during hypoxia breathing and not room air breathing (F). G: no change in hypoxic sensitivity (absolute value of the change in ventilation divided by the change in $\text{S}{\text{p}}_{{\text{O}}_2}$ measured during acute hypoxia and room air) was measured between normoxic and hyperoxic rats at P60. *P < 0.05 vs. age-matched normoxic group; ^P < 0.05 vs. within-group P10 value.

Given that the HVR was similar at all but one age (P21) between room air and hyperoxic rats, we adopted an alternative index of hypoxic sensitivity comparing the absolute value of the change in ventilation (hypoxia – room air) divided by the change in $\text{S}{\text{p}}_{{\text{O}}_2}$ (hypoxia – room air) between normoxic and hyperoxic rats. Based on this calculation, hyperoxia-exposed rats had significantly lower hypoxic sensitivity at P12 and P14 compared with normoxic rats (P < 0.05; Fig. 4E). Hypoxic sensitivity was similar between hyperoxic and normoxic rats at all other ages during development (P < 0.05; Fig. 4E) and as adults (P < 0.05; Fig. 5G). Therefore, hypoxia sensitivity transiently decreased after perinatal hyperoxia exposure but thereafter recovered to control levels indicating potential time-dependent plasticity.

Hyperoxia Exposure Alters Tidal Volume and Frequency Variability

We also noted age-dependent changes in breath-to-breath variability in both V_T and the interbreath interval (IBI). Short-term (SD1) and long-term (SD2) variability of V_T and the IBI were measured using Poincare analyses (also see methods). Scatter plots for V_T and IBI variability during room air breathing are depicted in Figs. 6A and 7A, respectively. During both room air and acute hypoxia breathing, hyperoxia-exposed rats had greater variability (SD1 and SD2) in V_T at P10 and P60 compared with normoxic rats (P < 0.05; Fig. 6, B and C). Additionally, hyperoxic rats had lower SD1 and SD2 values during room air and acute hypoxia from P12 to P60 when compared with P10 values (P < 0.05), whereas normoxic rats had similar levels of variability across all ages (P > 0.05). Hyperoxic rats also had greater SD1 and SD2 IBI variability at P10 but lower IBI variability at P12–P14 during room air (P < 0.05; Fig. 7B) and acute hypoxic conditions (P < 0.05; Fig. 7C). Unlike normoxic rats, hyperoxic rats had lower SD1 and SD2 values at P12, P14, P17, and/or P21 when compared with P10 values (P < 0.05; Fig. 7, B and C). Thus, perinatal hyperoxia exposure led to age-dependent changes in breath-to-breath variability.

Fig. 6.

Tidal volume (V_T) variability assessed by Poincare analyses was significantly altered acutely and chronically after hyperoxia exposure. A: Poincare plots of every quantified breath per group were plotted for normoxic rats (top plots) and hyperoxic rats (bottom plots) while room air was breathed. B and C: hyperoxia exposure caused significant changes in short-term (SD1) and long-term (SD2) variability during room air (B) and acute hypoxia breathing (C). The developmental pattern of SD1 and SD2 was assessed by comparing all values within a group to respective postnatal day (P)10 values, resulting in significant differences between P10 values and subsequent ages of hyperoxic rats but not normoxic rats. Poincare plots are not shown for acute hypoxia. Points per normoxic plot: P10 = 15,999, P12 = 17,390, P14 = 15,467, P17 = 17,450, P21 = 12,372, and P60 = 4,049. Points per hyperoxic plot: P10 = 21,681, P12 = 32,051, P14 = 29,779, P17 = 21,069, P21 = 12,811, and P60 = 4,720. In the P10 hyperoxic Poincare plot, SD1 and SD2 lines represent examples and not actual values, and the diagonal line represents the line of identification. *P < 0.05 vs. age-matched normoxic group; ^P < 0.05 vs. within group P10 value.

Fig. 7.

Breathing frequency variability assessed by Poincare analyses of the interbreath interval (IBI) was significantly altered acutely and chronically after hyperoxia exposure. A: Poincare plots of every quantified breath per group were plotted for normoxic rats (top plots) and hyperoxic rats (bottom plots) while room air was breathed. B and C: hyperoxia exposure caused significant changes in short-term (SD1) and long-term (SD2) variability during room air (B) and acute hypoxia breathing (C). The developmental pattern of SD1 and SD2 was assessed by comparing all values within a group to respective postnatal day (P)10 values, resulting in different ages being significantly different for normoxic vs. hyperoxic rats (B and C). Poincare plots are not shown for acute hypoxia. Points per normoxic plot: P10 = 16,003, P12 = 16,680, P14 = 15,466, P17 = 16,508, P21 = 12,507, and P60 = 3,781. Points per hyperoxic plot: P10 = 22,637, P12 = 33,734, P14 = 28,669, P17 = 21,124, P21 = 13,737, and P60 = 5,145. *P < 0.05 vs, age-matched normoxic group; ^P < 0.05 vs. within group P10 value.

Hyperoxia Exposure Induced Histological Changes Within The Brainstem Nuclei Controlling Breathing

Given the major ventilatory changes measured throughout development and up through adulthood, we aimed to test whether there might be evidence that perinatal hyperoxia impacted the central respiratory network controlling breathing. Immunohistochemistry was used to assess potential anatomic changes in neuron number (NeuN) and/or the expression of a marker of astrocytes (GFAP) within multiple key respiratory nuclei in adults. The number of neurons within the nucleus of the solitary tract (NTS) was greater in hyperoxic rats compared with normoxic rats (P < 0.05). The number of neurons counted within all other respiratory nuclei, including the hypoglossal motor nucleus (HG), dorsal motor nucleus of the vagus (DMV), ventral respiratory column (VRC; includes the nucleus ambiguus and pre-Bötzinger complex), retrotrapezoid nucleus (RTN), and subnuclei of the raphe [raphe pallidus (RPa), obscurus (ROb), and magnus (RMG)] were unaltered (P > 0.05) between normoxic and hyperoxic rats (Fig. 8). In contrast, GFAP expression, as measured by percent positive signal per imaged field (see methods), was greater in most of respiratory nuclei analyzed in hyperoxia-exposed rats compared with controls (P < 0.05). Within the VRC there was greater GFAP staining in hyperoxic rats (P < 0.05). In addition, the ventrolateral medulla, which includes the RTN, and within the subnuclei of the medullary raphe (RPa, Rob, and RMg), GFAP staining in hyperoxic rats was also greater relative to normoxic rats (P < 0.05; Fig. 9). GFAP levels were increased in the HG (P < 0.05), whereas in the NTS and the DMV, GFAP was unaltered (P > 0.05).

Fig. 8.

Histological quantification of neuron counts within respiratory control nuclei assessed by counting the number of NeuN (neuron number)-positive cells within a defined region of interest. Neuron counts were normalized to the area of region analyzed. The no. of neurons was elevated in the nucleus of the solitary tract (NTS). However, the no. of neurons in the hypoglossal motor nucleus (HG), dorsal motor nucleus of the vagus (DMV), ventral respiratory column (VRC), retrotrapezoid nucleus (RTN), raphe pallidus (RPa) raphe, raphe magnus (RMg), and raphe obscurus (ROb) was not different between hyperoxic (Hx) and normoxic (Nx) rats. *P < 0.05 vs. age-matched Nx group. Scale bar, 100 µm.

Fig. 9.

Histological quantification of glial fibrillary acid protein (GFAP) expression within respiratory control nuclei assessed by %area fraction of positive signal per ×20 high-powered field following immunofluorescence labeling. GFAP expression was elevated across key respiratory control nuclei at postnatal day (P)60 in hyperoxia-exposed rats. Representative GFAP images are shown for normoxic (Nx; top) and hyperoxic (Hx) rats (bottom), and quantification of GFAP expression within each nucleus is shown under representative histologic images. GFAP expression in the hypoglossal motor nucleus (HG), ventral respiratory column (VRC), retrotrapezoid nucleus (RTN), raphe pallidus (RPa) raphe, raphe magnus (RMg), and raphe obscurus (ROb) was significantly greater in Hx rats compared with Nx rats. No differences in GFAP expression were measured within the nucleus of the solitary tract (NTS) and dorsal motor nucleus of the vagus (DMV). *P < 0.05 vs. age-matched Nx group. Scale bar, 100 µm.

DISCUSSION

In the present study, we used an established rat model of BPD induced by perinatal hyperoxia exposure (26, 52) to investigate the acute and chronic effects on the control of breathing. Our data indicate that chronic perinatal hyperoxia exposure 1) causes a BPD-like phenotype assessed by lung histology and pulmonary function metrics, 2) led to sustained increases in ventilation and age-dependent shifts in hypoxic sensitivity, ventilatory pattern, and variability, and 3) induces a glial expansion without major changes in neuronal numbers within several key brainstem nuclei that contribute to the control of breathing.

Chronic perinatal hyperoxia exposure led to a BPD phenotype that was sustained throughout the 60-day study and was associated with significant increases in minute ventilation acutely and chronically. This increase in ventilation likely reflects an increase in alveolar ventilation and not an increase in dead-space ventilation given that increases in minute ventilation were due at least in part to an increase in tidal volume. Increasing alveolar ventilation likely reflects a compensatory mechanism in response to reduced gas exchange surface area and alveolar development commonly observed among BPD patients (27). Indeed, hyperoxia exposure significantly reduced alveolar surface area and the number of alveoli measured at P10 through P60. Moreover, we measured similar $\text{S}{\text{p}}_{{\text{O}}_2}$values between hyperoxic and normoxic rats during room air breathing but a higher V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$ ratio from P12 through P60, suggesting that greater alveolar ventilation is needed to sustain a normal $\text{S}{\text{p}}_{{\text{O}}_2}$. This could reflect hyperventilation, which would, therefore, cause a decrease in $\text{P}{\text{a}}_{{\text{C}\text{O}}_2}$, which may have occurred during development. However, at P60, both $\text{P}{\text{a}}_{{\text{O}}_2}$ and $\text{P}{\text{a}}_{{\text{C}\text{O}}_2}$ are unchanged perhaps due to adaptations in metabolic rate. The contributions to changes in metabolic rate require further investigation. Thus, perinatal hyperoxia exposure led to sustained elevations in minute ventilation, which may be necessary to maintain blood gases after developmental arrest of the lung. However, $\text{S}{\text{p}}_{{\text{O}}_2}$ remained significantly lower during acute hypoxic challenges, indicating that the hyperventilation is insufficient to correct for lower $\text{S}{\text{p}}_{{\text{O}}_2}$. This could be due to impaired lungs and/or acutely reduced hypoxic sensitivity, although hypoxic sensitivity was unchanged at P17 when acute hypoxic $\text{S}{\text{p}}_{{\text{O}}_2}$ values remained lower. Nonetheless, these results suggest chronic adaptations in the control of breathing in response to pulmonary disease but do not necessarily rule out direct effects of hyperoxia on the control of breathing. Additionally, the V̇e/$\text{S}{\text{p}}_{{\text{O}}_2}$ values appear to converge between the two groups, with age further suggesting compensation within the respiratory network over time.

Despite using different models of chronic perinatal hyperoxia (3, 5, 18, 31), prior studies report impaired development of the carotid bodies and hypoxic sensitivity, consistent with clinical observations in BPD infants (10). Our data partially support these prior findings. When unadjusted for lung injury, our HVR measurements were similar between normoxic and hyperoxic rats, which was a similar finding reported by Ling et. al (30). However, when Ling et al. (30) corrected for lung injury and assured that similar levels of hypoxemia were met during the acute hypoxia challenge, hyperoxic-exposed rats had blunted hypoxic sensitivity as adults (hypoxic sensitivity was not measured at earlier ages). Herein we did not assure equivalent levels of hypoxemia between the two groups of rats during the acute hypoxia challenges. Instead, we normalized the change in ventilation (hypoxia – room air) to the change in $\text{S}{\text{p}}_{{\text{O}}_2}$ values (hypoxia – room air) and found a blunted hypoxic sensitivity acutely (P12–P14) but not chronically in hyperoxic-exposed rats. The sustained increases in ventilation during room air and acute hypoxia and return of hypoxic sensitivity after P14 may be mediated by recruitment of other peripheral chemoreceptors (aortic bodies), as occurs after bilateral carotid body denervation in newborn piglets (49) and changes to the central control of breathing as suggested after carotid body denervation in adult rats (39), or driven by an increase in central hypoxia sensitivity of astrocytes (21).

The latter possibility is supported by our findings of enhanced GFAP expression within six of eight key respiratory brainstem nuclei in hyperoxic rats. GFAP is a cytoskeletal filament expressed in astrocytes and is a marker used for astrogliosis (41). Whether the increase in GFAP expression indeed reflects a functional change in contribution to ventilation is unclear from our data. Alternatively, increased GFAP may reflect hyperoxia-related oxygen neurotoxicity and consequent reactive gliosis (41) or neuroplasticity that drives the age-dependent changes in ventilation as astrocytes participate in synaptic function and plasticity (50) or a combination of both changes. Because the number of neurons did not change in all but one brainstem nucleus, these rather global increases in GFAP could alter neuronal function in this model. The brainstem region uniquely affected in this model is the NTS, which showed increased neuronal counts and no change in GFAP expression. It is unclear by which mechanism greater neuronal counts result in the NTS, but this site appears to be unique with respect to vagus nerve denervation-induced neuronal proliferation in the dorsal vagal complex (13). Overall, the data show that perinatal hyperoxia exposure sufficient to induce BPD caused increases in ventilation and specific and sustained anatomic changes within the respiratory control network.

In addition to measuring changes in minute ventilation, frequency, and tidal volume, we assessed the effects of hyperoxia-induced BPD on ventilatory stability by applying Poincare analyses to assess variability in tidal volume and breathing frequency (interbreath interval) acutely and chronically. Although it is postulated that some variability is healthy and inherent in physiological control systems, too much or too little variation may be pathological (12), causing instability or rigidity in the system, respectively. Acutely after hyperoxia exposure, the ventilatory control system appears to be “rigid” based on the significantly decreased frequency variabilities measured at P12–P14 during both room air and acute hypoxia conditions. The proposed “rigidity” coincides with a known and naturally occurring critical window of respiratory development during which chemosensitivity transiently decreases (32) and a shift of excitatory and inhibitory neural systems within respiratory nuclei takes place (32–34, 55). Moreover, we recently reported significant changes in the expression of multiple neurochemicals within respiratory nuclei and a delay in the onset of the critical window following the same hyperoxia exposure protocol as used in this study (40). Together, these data indicate that hyperoxia-induced BPD may disrupt the control of breathing beyond what naturally occurs such that there is a greater attenuation in the response to acute stressors. These additional effects may underlie the increased risk for SIDS in BPD infants (54).

Limitations

Although our rodent model relied on a nonclinically relevant ${\text{FI}}_{{\text{O}}_{\text{2}}}$ (>90%), rodents, unlike humans, are born with a more mature antioxidant system and, thus to induce BPD, require a high level of oxygen exposure. Moreover, as discussed by Ambalavanan and Morty (1), hyperoxia-associated BPD rodent models are at best mimicking moderate BPD despite using such a high ${\text{FI}}_{{\text{O}}_{\text{2}}}$. Thus, although the high ${\text{FI}}_{{\text{O}}_{\text{2}}}$ per se is not clinically relevant, it is necessary to induce BPD-like phenotypes in the rodents and leads to similar increases in tidal volume and breathing frequency, as observed clinically (16). Finally, the goal of the study was to understand the impact of perinatal hyperoxia associated with BPD on the development of the control of breathing. Further studies need to identify the specific contributions of lung injury to the changes in the control of breathing.

Conclusions

Our study validates our hypothesis that neonatal hyperoxia exposure would induce BPD and acutely and chronically alter the control of breathing. Based on our results, we conclude that hyperoxia-associated BPD causes significant changes to the control of breathing, which may be centrally mediated by astrocytes in the brainstem. Moreover, hyperoxia-mediated BPD is associated with an increase in respiratory rigidity and less responsiveness to acute hypoxic stressors during a critical window of development, which may lead to greater risk of SIDS than in non-BPD infants. These findings suggest that BPD infants may have fundamentally different mechanisms controlling breathing throughout their life.

GRANTS

This work was supported by the Children’s Research Hospital of Wisconsin Research Institute, National Heart, Lung, and Blood Institute Grant R01-HL-122358 and the Parker B. Francis Foundation.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

G.C.M., M.T.T.W.-R., G.G.K., and M.R.H. conceived and designed research; G.C.M., S.A.-A., R.G., G.T., and T.M. performed experiments; G.C.M., S.A.-A., R.G., G.T., M.T.T.W.-R., G.G.K., and M.R.H. analyzed data; G.C.M., S.A.-A., G.T., M.T.T.W.-R., G.G.K., and M.R.H. interpreted results of experiments; G.C.M. and S.A.-A. prepared figures; G.C.M. and M.R.H. drafted manuscript; G.C.M., S.A.-A., M.T.T.W.-R., G.G.K., and M.R.H. edited and revised manuscript; G.C.M., S.A.-A., T.M., M.T.T.W.-R., G.G.K., and M.R.H. approved final version of manuscript.