Neonatal hyperoxia in mice triggers long-term cognitive deficits via impairments in cerebrovascular function and neurogenesis

Preterm birth is the leading cause of death in children under 5 years of age. Premature infants who receive life-saving oxygen therapy often develop bronchopulmonary dysplasia (BPD), a chronic lung disease. Infants with BPD are at a high risk of abnormal neurodevelopment, including motor and cognitive difficulties. While neural progenitor cells (NPCs) are crucial for proper brain development, it is unclear whether they play a role in BPD-associated neurodevelopmental deficits. Here, we show that hyperoxia-induced experimental BPD in newborn mice led to lifelong impairments in cerebrovascular structure and function as well as impairments in NPC self-renewal and neurogenesis. A neurosphere assay utilizing nonhuman primate preterm baboon NPCs confirmed impairment in NPC function. Moreover, gene expression profiling revealed that genes involved in cell proliferation, angiogenesis, vascular autoregulation, neuronal formation, and neurotransmission were dysregulated following neonatal hyperoxia. These impairments were associated with motor and cognitive decline in aging hyperoxia-exposed mice, reminiscent of deficits observed in patients with BPD. Together, our findings establish a relationship between BPD and abnormal neurodevelopmental outcomes and identify molecular and cellular players of neonatal brain injury that persist throughout adulthood that may be targeted for early intervention to aid this vulnerable patient population.


Supplemental Methods
Arterial blood oxygen saturation measurements. The oxygen saturation measurements were conducted at P14 and at 12-14-months of age. Adult mice had undergone behavioral testing (see behavioral methods) at 7 months of age. Both age groups were tested at rest; however, adult mice were also assessed immediately following treadmill exercise (see treadmill methods). Mice were anesthetized with isoflurane and the Mouse Ox Plus® oximeter sensor (Starr Life Sciences Corp.) was placed over the shaved, right thigh of each mouse to detect blood oxygen saturation of the femoral artery. Arterial blood oxygen saturation, heart rate (beats per minute), and respiratory rate (breaths per minute) were recorded for a minimum of 25 seconds using the Mouse Ox Plus® basic software.
Lung histopathology and image acquisition. Lungs were intratracheally inflation fixed with 10% buffered formalin, under 20 cm H2O pressure, for 5 minutes. Lungs were immersion fixed in 10% buffered formalin for 48 hours at room temperature and immersed in 70% ethanol for 24 hours at room temperature. Lungs were then paraffin-embedded, cut into 4 µm sections, and stained with hematoxylin and eosin at the University of Ottawa Histology Core Facility. To visualize lung structure, images (x20 magnification) were acquired on an Aperio CS2 slide scanner (Leica) using the Aperio eSlide Manager (Leica).
Metabolic treadmill test. The treadmill speed increased as follows: 0 cm/s for 5 minutes, 8cm/s for 5 min, 15cm/s for 5 min, 20 cm/s for 5 min, 25 cm/s for 5 min, then increased by 3 cm/s every 2 minutes. Mice remained on the treadmill until exhaustion, as determined by the mouse remaining on the shock bar (~0.5mA) for 10 seconds, the mouse being unable to run, or if the respiratory exchange ratio became greater than or equal to 1 or dramatically increased within 4-6 minutes and then reached a plateau. Oxygen consumption and carbon dioxide release were measured every 30 seconds using the Oxymax®/ comprehensive lab animal monitoring system (CLAMS) and software (Columbus Instruments).
MRI. Mice were 14 months of age at time of testing and had previously been assessed for behavior at 12 months of age (see behavior methods). MRI was conducted using a 7 Tesla General Electric/Agilent MR901 machine. Coronal, 2D fast spin echo T2-weighted brain images were acquired (TE = 25 ms, TR = 6000 ms, ETL = 8, bandwidth = 15.6 KHz, FOV = 2.5 cm, slice thickness = 0.5 mm, matrix = 265 x 256, scan time = 7 min). The size of the whole brain, lateral ventricles, and hippocampal regions were measured using Fiji. The total area of anatomically comparable sections was measured using the polygon tool and was multiplied by the thickness of each MRI slice.
Laser doppler flowmetry and analysis. The same cohorts of animals were tested at P14 and 10 months of age. Briefly, once the animal was mounted in the stereotactic equipment and the skull was exposed, the flowmeter (BLF22, single-channel tissue perfusion monitor, Transonic Systems Inc.) was placed above the left somatosensory cortex. Baseline CBF measurements were recorded. To assess neurovascular coupling, whiskers on the right side of the animal were stimulated for 20 second intervals a minimum of 3 times. An average of three stimuli was used for the baseline and whisker stimulated CBF values for each animal.
Systolic blood pressure. Systolic blood pressure was measured 5 times, to ensure consistency, followed by 10 trials which were recorded. Training occurred over 5 consecutive days. After a 2day interval, systolic blood pressure was measured on the acclimatized mice over 5 consecutive days, following the same procedure as during the training sessions. Only consistent measurements of at least 8 of the 10 trials, per day, were used in the analysis. The average systolic blood pressure of approximately 50 trials per mouse was calculated.
NPC niche region image analysis. All image analysis was conducted while blinded to the experimental groups. 3D modelling of z-stack images was conducted using the surfaces module of Imaris 9.3 to measure the volume of i) the SVZ and ii) the SGZ (for NPC quantification) or DG (for newborn neuron quantification) of sections. For P14 mice, a total sum of 2 sections per region, per animal was used in the analysis. For 12-month-old mice, a total sum of 3 sections per region, per animal was used in the analysis. For NPC quantification, computational analysis of the SGZ was conducted with Imaris 9.3 (Bitplane Inc.) using the spots module to quantify NPCs (Sox2 + , nestin + ) and confirmed with manual inspection. For the SVZ, images were manually quantified using the cell counter module of Fiji. The contact points of CD31 + ECs with nestin processes of NPCs in the DG was quantified manually using the cell counter module of Fiji. Newborn neurons (DCX + ) were manually quantified in the SVZ and DG using the cell counter module of Fiji. Images were processed in Fiji (adjustments applied equally to all images within a comparison) and are displayed as maximum intensity Z-projections.  water and mounted with Fluoromount-G or ProLongGold with DAPI. Sections were imaged as 8 μm z-stacks and were acquired (x20 magnification) on a Zeiss Axio Imager.M2 with an ApoTome.2 system. Images were manually quantified using the cell counter module of Fiji.
Images were processed in Fiji (adjustments applied equally to all images within a comparison) and are displayed as a maximum intensity Z-projection. A total sum of two sections per region per animal was used in the analysis. A minimum of 200 cells/ section from 3 fields of view were counted. The cells expressing 2 or more identifiers of the RNA of interest, Ctla2a, were counted as positive. A percentage of Ctla2a positive cells was calculated from the total cells counted.
NPC subpopulation image analysis. Images were manually quantified using the cell counter module of Fiji. The area of each counted region was measured using the polygon tool of Fiji.
Images were processed in Fiji (adjustments applied equally to all images within a comparison) and are displayed as a maximum intensity Z-projection.
Behavioral experiments. Adult mice at 7 months and 12 months of age were handled for 2 days before the experiments commenced and the person conducting the testing was blinded to the treatment groups. All testing was conducted during the dark cycle and in red light, with the mice being habituated to the red-light testing room for ~30-60 minutes prior to testing, unless otherwise indicated.
Rotarod. Mice had four trials (10-minute inter-trial interval (ITI)) over a period of 2 consecutive days on an accelerating rotarod (IITC Life Science Inc.). For each trial, the rod was set to accelerate from 4 to 45 rpm in 300 seconds, followed by 300 seconds at 45 rpm.
DigiGait™. Mouse gait was recorded for a minimum of 3 seconds on the transparent DigiGait™ treadmill (Mouse Specifics, Inc.) that was set to a speed of 18 cm/sec, and an incline of 8 degrees.
Home cage locomotor activity. Mice were placed individually into clean housing cages which were then placed for 4 hours into the Home Cage Locomotor Activity Infrared Beam Break frames that were paired with the Fusion software (Omnitech Electronics, Inc.).
Morris water maze. Each mouse was habituated to the testing room in 140 lux light for 30 minutes.
A single black X (2.8 cm thick, 15cm long x 13.5cm wide) was placed on the back wall of the room as a cue. Each mouse was placed in a circular pool, measuring 132 cm in diameter, filled with water colored white with tempera paint maintained at 23°C. The pool contained a hidden platform (10 cm in diameter) in one of its 4 quadrants. Each mouse was trained to find the platform for four trials each day (ITI of 30 minutes) for nine days total. During each trial, the mouse had 1 minute to find the platform. On the 10 th day, the platform was removed, and the mouse was given one minute to search for the platform. Each trial was tracked and analyzed using Ethovision software (Noldus). If the mouse did not find the platform within 1 minute, the program automatically stopped.
Fear conditioning. Mice were not habituated to the testing room prior to the experiment and the experiment was performed using the PhenoTyper® boxes (Noldus) with grid shock floors (Med Associates). The overhead room lights were on and no lights were projected by the boxes. On the training day, mice were placed into the box. The mice remained in the box for 2 minutes, after which a 30 second 80 dB tone played, followed by a 2 second 0.45mA foot shock. This tone-shock pairing was repeated two times with a 1-minute interval. The final tone-shock pair was followed by a 30-second interval. The mice were then returned to their home cages. On day 2, to assess contextual memory, mice were placed in the same box for the same duration as training with no tone or shock. On day 3, to assess cued memory, mice were placed in a different Phenotyper® box from the one used for training and the contextual memory test. This box was altered through the addition of a plastic floor, triangular plastic walls, and a vanilla scent. The room was lit with red light and the boxes were lit with both white and yellow light simultaneously. The mice were in the modified box for six minutes with the 80 dB tone playing during the last three minutes, in the absence of any shock. All trials were recorded and freezing behavior was scored using Ethovision software (Noldus). Mice that froze less than 10 seconds were not included in the analysis.
Anesthesia was maintained throughout the test and the medetomidine was reversed after 1 hour using atipamezole hydrochloride (1 mg kg −1 ; Antisedan®). Eyes were dilated for 10 minutes prior to ERG testing with one drop each of 1% tropicamide (Mydriacyl, Alcon) and 2.5% phenylephrine hydrochloride (Mydfrin, Alcon). A topical anesthetic (0.5% proparacaine hydrochloride; Alcain, Alcon) was applied to each eye. 1 mL of saline was administered subcutaneously, prior to testing. Ag/AgCl contact stimulators were placed on both corneas in combination with Optixcare Veterinary Eye Lube (Aventix) to ensure stimulator contact and corneal hydration. A gold loop reference electrode was placed on the tongue, a needle electrode was placed sub-dermally in the head, and a grounding needle electrode was placed subcutaneously in the tail. Retinal function was assessed with the following 3 protocols. The simultaneous ERG Fundus photography. Mice were anesthetized, dilated, and maintained as described in the ERG protocol above. Fundus imaging was acquired using Streampix 3 (Norpix) on a Micron III microscope (Phoenix Technology Group) to inspect retinal morphology. Figure 1. Early life hyperoxia does not influence arterial oxygen saturation, oxygen consumption, or carbon dioxide production in adulthood. (A) Three outcome measures were assessed for 12-14-month-old mice at rest. From left to right: blood oxygen saturation of the femoral artery; heart rate (beats per minute, bpm); and respiratory rate (breaths per minute, brpm) (normoxia, n=20; hyperoxia, n=17; two-way ANOVA with Sidak post hoc test for group comparisons). (B) Volume of oxygen consumption, volume of carbon dioxide release, and the respiratory exchange ratio (VCO2 / VO2) for the total duration of metabolic treadmill testing (top) and at maximum capacity (bottom) (normoxia, n=20; hyperoxia, n=17; two-way ANOVA with Sidak post hoc test for multiple comparisons and unpaired Student's t test, respectively). Mice were tested at 12-14 months of age. Data are expressed as mean ± SEM.   Figure 2. Delayed brain growth after early developmental hyperoxia exposure.