Abstract
Prematurely born infants often experience frequent hypoxic episodes due to immaturity of respiratory control resulting in disturbances of cortical development and long-term cognitive and behavioral abnormalities. We hypothesize that neonatal intermittent hypoxia alters maturation of cortical excitatory and inhibitory circuits that can be detected early with functional MRI. C57BL/6 mouse male and female pups were exposed to an intermittent hypoxia (IH) regimen from P3 to P7, corresponding to preterm humans. Adult mice after neonatal IH exhibited motor hyperactivity and impaired motor learning in complex wheel tests. Patch-clamp and evoked field potential recordings revealed increased glutamatergic synaptic transmission. To investigate the role of GABAergic inhibition on glutamatergic transmission during the developmental, we applied a selective GABAA receptor inhibitor picrotoxin. A decreased synaptic inhibitory drive in the motor cortex was evidenced by miniature IPSC frequency on pyramidal cells, multi-unit activity recording in vivo with picrotoxin injection, and decreased interneuron density. There was also an increased tonic depolarizing effect of picrotoxin after IH on Betz cells’ membrane potential on patch-clamp and direct current potential in extracellular recordings. The amplitude of low-frequency fluctuation on resting-state fMRI was larger, with a larger increase in regional homogeneity index after picrotoxin injection in the IH group. The increased glutamatergic transmission, decreased numbers, and activity of inhibitory interneurons after neonatal IH may affect the maturation of connectivity in cortical networks, resulting in long-term cognitive and behavioral changes. Functional MRI reveals increased intrinsic connectivity in the sensorimotor cortex, suggesting neuronal dysfunction in cortical maturation after neonatal IH.
Significance Statement
The study demonstrates that perinatal hypoxic brain injury disrupts the balance between excitatory and inhibitory neurotransmission in developing cortical networks. This disruption, potentially caused by functional deficiencies in GABAergic interneurons alongside increased glutamatergic transmission, may contribute to altered brain connectivity and the observed behavioral deficits, including hyperactivity and cognitive difficulties. This research provides insights into how perinatal brain injury disrupts the balance of neural excitation and inhibition, which can be detected as altered local resting-state fMRI connectivity. These findings contribute to our understanding of possible cellular underpinning of clinical fMRI findings after perinatal brain injury.
Introduction
Despite advancements in neonatal care, perinatal hypoxic brain injury remains a significant risk factor for motor, behavioral, and cognitive impairments (Dilenge et al., 2001; van Handel et al., 2007; Pascal et al., 2018). Intermittent hypoxia (IH) is a common condition in preterm infants due to immature breathing control (Eichenwald, 2016), contributing to impaired neurodevelopment (Pillekamp et al., 2007; Poets, 2019). Preterm infants, especially those born very prematurely (<32 weeks gestation), are particularly susceptible to hypoxic brain injury and often exhibit deficits in memory, attention, and executive function (Ketzer et al., 2012; Burnett et al., 2013; Anderson, 2014; Johnson and Marlow, 2017; Mikkelsen et al., 2017).
Animal models of perinatal hypoxia replicate many human neurodevelopmental impairments, including learning, memory, attention, and motor deficits (Sab et al., 2013; Maxwell et al., 2020). Perinatal hypoxia likely disrupts the delicate balance between excitatory and inhibitory brain circuits. Previous studies indicate that perinatal hypoxia reduces major interneuron populations (Komitova et al., 2013; Chavez-Valdez et al., 2018; Stolp et al., 2019; Yang et al., 2023), but the impact on network activity and glutamate signaling remains unclear. Mice exposed to neonatal hypoxia exhibit increased neuronal activity and are prone to seizures (Burnsed et al., 2019). This elevated excitability has been attributed to increased AMPA receptor-mediated synaptic transmission (Jensen et al., 1998; Lippman-Bell et al., 2016). In a mouse model of neonatal IH, we also observed abnormalities in glutamatergic transmission, including elevated NMDA and AMPA receptor-mediated synaptic currents, along with a reduction in the number of AMPAR-lacking “silent” synapses (Goussakov et al., 2021). Disruptions in the excitatory/inhibitory balance, caused by increased excitatory transmission and decreased interneuron function after perinatal hypoxia, may impair synaptic pruning and neuronal circuit formation during this critical developmental period (Rakic et al., 1986). This can lead to hyperconnected networks (Kerchner and Nicoll, 2008), potentially contributing to cognitive, behavioral, and memory deficits. Additionally, the maturation of principal cells and interneurons is influenced by the shift of GABAergic transmission from excitatory to inhibitory during early development (Sauer and Bartos, 2010; Ben-Ari et al., 2012). A critical question remains whether reduced interneuron function adequately counteracts heightened excitatory signaling after perinatal hypoxia.
We hypothesized that injury to interneurons after neonatal global hypoxia may disrupt cortical maturation of the excitatory/inhibitory balance in the primary motor cortex (principal excitatory) neurons (Betz cells), involving both synaptic and extrasynaptic, phasic and tonic, respectively, modulation of glutamatergic transmission. In this study, we focused on the motor cortex because sensory-motor abnormalities are a major clinical manifestation of perinatal brain injury, including impairments in motor learning and voluntary control (Cahill-Rowley and Rose, 2014), as well as motor hyperexcitability and hyperactivity (van Handel, Swaab et al., 2007). To bridge the gap between in vivo fMRI findings and underlying cellular mechanisms, the study utilized ex vivo brain slice preparations and invasive in vivo recordings to investigate the impact of perinatal brain injury on neural circuits.
Materials and Methods
The study has received approval from the Institutional Animal Care and Use Committees of NorthShore University Health System.
Neonatal intermittent hypoxia
C57BL/6J mice were originally obtained from the Jackson Laboratory and bred at the NorthShore University Health System animal facility. Neonatal mice of both genders were randomly assigned to the IH or control groups. Animals underwent an intermittent hypoxia regimen using cycles of hypoxia with a 5% O2/95% N2 gas mixture for 2.5 min, followed by 5 min of normoxia with room air (Goussakov et al., 2019). During hypoxic exposure, pups were separated from their dams and placed into a 250 ml airtight chamber in a temperature-controlled neonatal incubator with an ambient temperature of 35°C. The paradigm began at postnatal day 3 (P3) and continued for 5 consecutive days. Airflow was maintained at 4 L/min and switched between the hypoxic mixture and room air using solenoid valves controlled by a programmable timer. Oxygen concentration and air temperature within the chamber were continuously monitored with calibrated sensors, reaching a nadir of 5% O2 within 40 s during hypoxic periods. The total number of daily hypoxic episodes gradually decreased across the 5 d (20, 18, 11, 9, and 8), divided into morning and afternoon sessions. Between the sessions, pups were returned to their dams for 4 h for recovery and nursing. Control mice underwent similar handling procedures, including separation from mothers, but remained in room air inside the incubator throughout the experiment. Mortality after neonatal IH was 28.5%. The number of animals is indicated for each outcome measure below.
Motor cortex slice preparation and patch-clamp recording of Betz cells
Mice between P7–P13 and P50–P60 were deeply anesthetized with isoflurane (3%) and decapitated. Brains were rapidly removed in ice-cold artificial cerebrospinal fluid (aCSF) solution and the 300-µm-thick slices, containing M1 motor cortex, were obtained at the level of forelimbs according to Paxinos atlas (Paxinos and Franklin, 2004) using a Leica Vibratome (VT1000S, Leica Biosystems). The aCSF solution for slice preparation consisted of the following reagents (in mM): 68 NMDG, 2.5 KCl, 1.25 NaH2PO4, 30 NaHCO3, 20 HEPES, 2 thiourea, 25 d-glucose, 3 Na-ascorbate, 0.3 ml/L ethyl pyruvate, 0.5 CaCl2-2H2O, 10 MgSO4-7H2O, 2 kynurenic acid. The final pH was adjusted to 7.3 with 1N HCl, osmolality was 300 ± 5 mOsm/kg H2O. The NMDG was used for slicing as a substitute for NaCl, to increase cell viability in slice preparation. Slices were transferred to a custom-made incubation chamber with a storage solution, containing the following (in mM): 120 NaCl, 2.5 KCl, 1.25 NaH2PO4, 25NaHCO3, 10 d-glucose, 2 thiourea, 3Na-ascorbate, 0.3 ml/L Na-pyruvate, 2 CaCl2-4H2O, 2 MgSO4-7H2O, 2 kynurenic acid, and 10 HEPES/NaOH, pH 7.3, and osmolality 300 ± 5 mOsm. Slices were kept on a nylon mesh and bubbled with carbogen (95% O2/5% CO2) to maintain a pH of 7.4 for at least 1 h before recording. For patch-clamp recordings, slices were transferred to a submerged chamber (model RC-24, Warner Instruments) and perfused at a rate 3 ml/min at room temperature with aCSF of the following composition (in mM): 120 NaCl, 2.5 KCl, 25 NaHCO3, 2 CaCl2, 2 MgSO4, 1.25 NaH2PO4, 10 HEPES, 10 d-glucose, bubbled with carbogen. Electrodes were pulled from capillary tubes 1B150F-4 (World Precision Instruments) using P97 puller (Sutter Instrument) and filled with a solution composed of the following (in mM): 110 K-gluconate, 15 Cs-methanesulfonate, 10 TEA-Cl, 5 QX-314, 4 NaCl, 2 MgCl2, 10 EGTA, 10 HEPES, 2 MgATP, 0.3 NaGTP for miniature currents recording in whole-cell configuration. The solution was titrated with CsOH to pH 7.2 with an osmolality of 290 ± 5 mOsm. Filled electrodes had a resistance of 2–5 MΏ.
Betz cells, identified by 30–50 μm size and location in layer 5 of the primary motor cortex, were recorded in whole-cell configuration. Series resistance (Rs) was not compensated but monitored and recordings with >15% change in Rs were discarded. Postsynaptic currents (PSCs) were amplified using an Axopatch-700B amplifier, digitized at the sampling rate of 100 kHz, and recorded in a voltage clamp using pClamp-10 software (Molecular Devices). Liquid junction potential was calculated using a tool in pClamp10 as −9.8 mV and resting membrane potential (Vm) was corrected in data analysis. Spontaneous miniature currents events were isolated by blocking action potentials with 1 μm tetrodotoxin (TTX). Glutamatergic miniature inwardly directed excitatory postsynaptic currents (EPSCs) were recorded at a holding potential of −70 mV [n = 10, (27 cells) in control and n = 12 IH (31 cells)]. After 10 min recording time, the holding potential was changed to 0 mV, and spontaneous GABAergic inhibitory postsynaptic currents (IPSCs) were recorded for another 10 min as outwardly directed currents. The polarity switch sequence was altered randomly between recordings. In control experiments, the glutamatergic nature of EPSCs was confirmed by a complete block of spontaneous activity with 6-cyano-7-nitroquinoxaline-2,3-dione disodium salt (CNQX, 50 μM, Sigma-Aldridge) and d-2-amino-5-posphonopentanoic acid (APV, 100 μM; Tocris) at a holding potential −70 mV (n = 12). Spontaneous IPSCs were identified as GABAergic by their suppression with 100 μM picrotoxin (Tocris) at 0 mV holding potential. PSCs were analyzed with MiniAnalysis software (version 6.0.3 (Synaptosoft) by manual selection events over a 2.5 pA threshold. At least 200 of each PSC's per cell was selected for analysis. Mean amplitudes and frequency of PSC were calculated.
Miniature IPSCs (mIPSCs) were chosen to examine the intrinsic properties and number of individual synapses to characterize baseline synaptic activity, mirroring the conditions during resting-state fMRI acquisition. mIPSCs examination allows evaluation of amplitudes and frequencies of the synaptic currents and is particularly suitable in the motor cortex where the source of inhibitory input is difficult to identify unambiguously.
Tonic GABAA receptor (GABAAR) inhibition was explored by monitoring Vm in whole-cell configuration with the addition of a specific inhibitor picrotoxin to evaluate the effect of extrasynaptic GABAARs [n = 6, 5 (16 and 18 cells) in control and IH for P10–P13, n = 20, 5 (32 and 11 cells) in control and IH for adults]. After 20 min of stable baseline, the Vm was measured during 20 min perfusion of 200 µM picrotoxin. The patch pipette was filled with the same solution as for miniature current recordings, except that Cs was replaced by equimolar concentration of K-gluconate to avoid potential effect on GABAARs. The offset of pipette potential was applied immediately after penetration of the solution surface in the recording chamber, and Vm was monitored in the current-clamp mode. The recordings were accepted if the change of continuously measured access resistance was <15%.
Field potential recording and analysis
Brain slices from the primary motor cortex were prepared as above. Field potential recordings were performed at several ages in neonatal and adult mice. After incubation for at least 1 h, the slices were transferred to the submerged recording chamber model RC-24 (Warner Instruments) and perfused at a 3 ml/min rate by gravity. At this rate, the 500 µl volume of the recording chamber is refilled in 10 s. A glass recording electrode, filled with aCSF, was placed in the middle of the molecular layer of the M1 cortex. A bipolar stimulation electrode, composed of tightly twisted 25-µm-diameter platinum/iridium wires, was placed 200–400 µm away from the recording site in the middle of the molecular layer. The distance between the electrodes was adjusted to avoid pop spikes that may otherwise introduce secondary activation by firing stimulated afferent cells at all ranges of stimulation intensity 0–700 µA. Input–output curves were constructed by plotting the amplitude as a function of the stimulation intensity. All recordings were obtained using a patch-clamp amplifier MultiClamp 700B (Molecular Devices) with a low-pass filter disabled to measure direct current potential (DCP) from the sweep baseline. Pipette potential offset was applied immediately in the solution before penetrating the slice preparation. The high-pass filter of the amplifier was set to 3 kHz for better resolution of the initial slope, and the Digidata 1442 was used to sample data points at 100 kHz and with pClamp 10.1 software (Molecular Devices).
A 100 µs duration monopolar stimulus pulses were triggered from Digidata 1442 using Clampex 10.1 software and delivered with a stimulus isolator A360 (World Precision Instruments). The field excitatory postsynaptic potentials (fEPSPs) were recorded in normal aCSF in a range of stimulus intensities (0–700 µA), with a step of 50 µA, and triggered once per minute. Three responses per intensity were used for averages. fEPSP amplitude baseline was recorded at 500 µA stimulus intensity in normal aCSF. The baseline was recorded for at least 20 min until stabilization of fEPSP's amplitude was achieved for at least 20 min. Then, 200 µM picrotoxin was added to perfusate and the recording continued for at least 20 min. The stimulus intensity of 500 µA was chosen to evoke a reliable fEPSP in the neonatal cortex without a pop spike appearance. After several pilot recordings, the picrotoxin concentration was selected to produce a substantial effect in neonatal mice cortex without invoking seizure activity. The swipes were recorded in Clampex for all time between triggering stimulus isolator to monitor possible seizure activity during picrotoxin perfusion. The initial slope of fEPSP was measured within 4 ms from the beginning of field potential. The time course of the picrotoxin effect on the fEPSP slope was calculated as a percentage over the average baseline, recorded in normal aCSF. DCP was measured as the averaged prestimulus sweeps baseline magnitude to evaluate tonic GABAAR-mediated inhibition with the addition of picrotoxin.
A total of 2–4 slices was recorded per animal. The number of animals was n = 4, 6, 7, 7, 5, 7, 6, 6 for controls and n = 3, 4, 5, 5, 5, 4, 4 for IH at ages P7, P8, P9, P10, P11, P12, P13, and adults, respectively. Only male mice were used to reduce variability in measurements. The recordings were averaged per slice and then averaged per each animal for final comparison.
Animal instrumentation for in vivo MRI and electrophysiological recording
At P11, mouse pups were instrumented with a custom 3D-printed circular head post 10 mm diameter with an opening 6 mm in diameter and with a small height profile to minimize disturbances to the dam while nursing in the nest. During MR imaging and electrophysiological recording, the head post was coupled with an adapter allowing head fixation to the animal cradle. The head post was attached to the skull using acrylic adhesive and dental cement. A small cranial window of ∼2 × 2 mm was made using a 25 ga needle above the right motor cortex with coordinates AP 1.5, ML 1.0. A plastic cannula and electrode guides were positioned above the cortex, and the opening was covered with dental cement. Animals received post-op analgesic Buprenex 0.1 mg/kg and left to recover at least 24 h before imaging and electrophysiological recordings.
Extracellular multi-unit recording in awake neonatal mice
P13–P14 mice (n = 5 for controls and n = 5 for IH, both sexes) were sedated with 1.5–2% isoflurane and placed in a custom cradle with the head secured using a head post adapter. The animal was lightly swaddled in gauze to maintain body temperature and minimize movement. The cradle was positioned in a stereotaxic device inside the Faraday cage and the anesthesia stopped. The recordings started after a 20 min waiting period to ensure the animal was fully awake (Aksenov et al., 2018). A 12 MΩ tungsten electrode (A-M Systems, catalog #577200) was lowered through the implanted guide cannula targeting layer 4 of the motor cortex at stereotaxic coordinates AP 1.5, ML 1. A ground wire was placed subcutaneously in the neck region. The electrophysiological signal was amplified and filtered through a miniature preamplifier connected to a differential amplifier (NeuraLynx). The signals were further amplified, bandpass filtered at 300 Hz–3 kHz to register multi-unit activity (MUA), and digitized at 32 kHz per channel using the NeuraLynx data acquisition system. Offline MUA analysis was performed using threshold detection followed by waveform analysis to exclude noise, using SpikeSort 3D software (NeuraLynx). The threshold criteria were set to three times the estimated standard deviation of the background noise (Aksenov et al., 2023). Following a 9 min baseline recording, 1 µl of 0.3 mM picrotoxin solution in aCSF was superfused through the second cannula guide above the recording area. This specific concentration of picrotoxin was chosen based on previous and preliminary studies demonstrating sufficient changes in neuronal activity without inducing epileptiform activity in the recordings of healthy animals (Aksenov et al., 2023). Control animals received an equal volume of vehicle (aCSF) using the same procedure to account for potential injection-related effects. An additional 30 min of electrophysiological recordings were obtained following picrotoxin or aCSF injection.
First, we compared MUA values before and after picrotoxin/aCSF injection within each experimental group. Second, we compared MUA values between the control and IH groups. Since the peak effect of picrotoxin might occur at slightly different times due to variations in drug diffusion across animals, we segmented the postinjection data into 5 min intervals. We then exclusively utilized the 5 min interval within each animal that exhibited the maximum change in MUA for statistical comparisons. To assess changes within each group induced by the injection, we utilized a paired two-tailed t test. A two-tailed, two-sample unequal variance t test was used for comparisons between the groups.
Resting-state functional MRI with drug injections
Resting-state functional MRI was performed on a 9.4T Bruker BioSpec scanner (Bruker) in mouse pups at P12 (n = 8, 8 for controls and IH, both sexes, 4 males and 4 females for each group). Mice were sedated with isoflurane (Abbott) inhalation, diluted in air to 3% for induction and 1.0% for maintenance. The animal’s respiration rate and rectal temperature were monitored with a small animal physiological monitor (Model 1030, Small Animal Instruments). Body temperature was monitored using a rectal probe and kept at 35°C by blowing warm air. Animals were placed prone in a cradle with their heads secured beneath the surface coil. The cradles were inserted into the scanner bore and aligned against a positional stopper. The setup allowed cradle retraction for injections and repositioning in the magnetic center with an accuracy of ±0.3 mm along the scanner's z-axis. The receiver coil was a 16 mm surface coil (Doty Scientific) allowing full mouse brain coverage. The transmitter was a 70 mm quadrature volume coil. Anatomical reference was acquired using RARE sequence TR/TE/NEX 3800/14/2 covering the cerebral cortex. The anesthesia was discontinued for at least 30 min to allow the animals to wake up before functional imaging. Resting-state functional MRI (rsfMRI) time series was acquired with the same geometry as the reference image using single-shot gradient echo T2*-weighted EPI, TE 18 ms, temporal resolution 1.5 s between volumes, 400 volumes. The field of view was 2.0 × 1.5 cm, matrix 128 × 96, in-plane resolution 0.156 × 0.156 mm2, and slice thickness 0.5 mm. After the first fMRI run, the cradle was retracted and 1 µl of 0.3 mM picrotoxin was slowly injected at the rate of 0.2 µl/min into the motor cortex through the cannula guide on the depth 1 mm using a microinjector and Hamilton syringe with 33 ga needle and left in the tissue for additional 1 min before slow retraction. The injection site was marked with a small bead of petroleum jelly to aid localization on MR images. The cradle was reinserted inside the scanner against the positional stopper. A new reference scan was acquired, followed by the second 10 min rsfMRI run.
Postprocessing was done with custom scripts in Matlab. Brain volumes were motion corrected using SPM12 and 9 degrees-of-freedom affine transformation. Severe motion outliers were detected and removed using the ArtRepair toolbox for SPM (https://www.nitrc.org/projects/art_repair). No spatial filtering was applied. Linear regression analysis was performed to remove sources of spurious variance using the following nuisance variables: six parameters obtained by rigid body analysis of head motion and a signal from cerebrospinal fluid regions of interest located in the lateral ventricle.
Maps of Amplitude of Low-Frequency Fluctuations (ALFF), fractional ALFF (fALFF), and regional homogeneity (ReHo) were calculated with a Matlab routine modified from the REST toolkit (https://www.nitrc.org/projects/rest/). AALFF and ReHo were chosen as two commonly used metrics in rsfMRI to assess spontaneous brain activity. AALF is also sensitive to changes in neural activity, while ReHo is more sensitive to local changes in neural synchronization. While AALF and ReHo provide complementary information, combining them can offer a more comprehensive understanding of brain function. The signal in each voxel was normalized to a global mean (Xi et al., 2012). After transforming voxel time series frequency information into the power domain, ALFF was calculated as the sum of amplitudes within a low-frequency range (0.008–0.3 Hz). fALFF was intended to reduce the sensitivity of ALFF to physiological noise by taking the ratio of the power spectrum in each frequency band (0.008–0.1 Hz (0.1–0.3 Hz) to the total frequency range (0–0.3 Hz). ReHo (Zang et al., 2004)was defined as the connectivity of a given voxel to those of its nearest 19 neighboring voxels using Kendall’s coefficient of concordance. A 1.5 × 1 mm oval region of interest (ROI) was placed around the injection site. Mean values of the above rsfMRI measures were extracted from the fMRI runs before and after the picrotoxin injection. The analysis was conducted in the native imaging space.
Behavioral assessment
All behavioral testing procedures were conducted in a dedicated quiet room between 41 and 50 d of age during the daytime. Number of animals was 6 males and 7 females for controls and 11 males and 12 females for IH. Each apparatus was thoroughly cleaned with 70% ethanol and aired for 3 min between animals. Animal movements were tracked and analyzed using ANY-maze software version 7.4 (Stoelting).
Open field test
Mice were placed individually in the center of a clear open field box (61 × 61 cm) and allowed to explore freely for 20 min. The software automatically recorded the following parameters of spontaneous locomotor activity: total mobile time, mean speed, total distance traveled, and time spent in the center zone (40 × 40 cm), compared with the peripheral zone. The center zone time is often used as an indicator of anxiety-like behavior in rodents.
Accelerating rotarod test
This test assessed balance, motor coordination, and motor learning. Mice were placed on a horizontal rotating rod set at 4 rpm (Harvard Apparatus). The rod's speed gradually accelerated from 4 to 40 rpm over a 5 min trial. The latency to fall (time spent on the rod) was recorded. Each mouse received four trials per day for 5 consecutive days, with 20–60 min of rest in their home cage between trials.
Complex wheel test
A training wheel with full 38 rungs (Lafayette Instrument) was presented individually for 3 consecutive days during day time allowing for normalization of running behavior. On the fourth day, the training wheel was replaced with a complex wheel of the same diameter but with 22 rungs missing in an alternating pattern. Running activity was recorded for an additional 4 d. Wheel speed was measured using an optical sensor and custom software with a 1 s resolution. The session duration was adjusted to limit of active running time for each animal to 1 h/d. The following outcome measures were calculated to assess motor performance and motor learning: total distance traveled, maximum and average speed, duration of running episodes, and average variability in running speed (coefficient of variation) in 10 s bins.
RNA isolation and real-time PCR
Mouse brains (n = 9, 11 for control and IH) were collected, and motor cortex was separated and individually frozen on liquid nitrogen and stored at −80°C before use. The tissue was homogenized in QIAzol Lysis Reagent (catalog #79306, Qiagen) on ice to extract total RNA. NanoDrop (Thermo Fisher Scientific) instrument was used to measure RNA quantity and quality. Then, 500 ng of isolated total RNA was used to synthesize cDNA using an RT2 First Strand Kit from QIAGEN. cDNA was amplified by polymerase chain reaction (PCR) with QuantiTect Sybr Green PCR Kit (Qiagen) on an Applied Biosystems QuantStudio 7 Flex real-time quantitative PCR instrument. The total reaction volume of 25 μl consists of 1.0 µl RT product (cDNA). Primers for Gria1-4, Gad1, Gad2, Slc12a2, and Slc12a5 GAPDH for Sybr Green Assay were predesigned and ordered from Integrated DNA Technologies. For each primer, the final concentration was 0.4 μm and the annealing temperature was adjusted. Gene expression was normalized to the housekeeping gene GAPDH and expressed as a fold change of experimental controls ΔΔCt method.
Stereological assessment of interneuron subpopulations
Mice were killed at P12 (n = 5, 5 for control and IH) and P45–P60 (n = 5, 5 for control and IH) by transcardial perfusion with saline solution (0.9% NaCl) followed by 4% paraformaldehyde (PFA). Brains were removed and postfixed in 4% PFA overnight, soaked in 30% sucrose solution for cryoprotection, frozen on dry ice, and cryosectioned 20 μm in the coronal plane. Primary antibodies against parvalbumin (ab181086, Abcam), and somatostatin (ab111912, Abcam) were used correspondingly. This was followed by incubation with biotinylated secondary goat anti-rabbit IgG antibodies (1:200; VectorLabs) for 1 h at room temperature and Avidin biotin complex for 1 h. The color was developed using 3,3′-diaminobenzidine (Sigma-Aldrich). Stereological estimation of neuronal cell density was performed under a microscope (Leica Microsystems) attached to a motorized stage (Ludl Electronic Products). Optical fractionator probes were used to estimate the number of labeled cells in the motor cortex in Stereo Investigator software (MBF Bioscience). Six 20-μm-thick coronal sections 300 μm apart, starting ∼1 mm from the end of the olfactory bulbs and covering primary and secondary motor cortex, were counted with a counting frame 350 × 350 μm2 and a sampling grid of 1,100 × 1,100 μm2. The coefficient of error of the stereological estimation (Gundersen) for each animal ranged from 0.05 to 0.1. Cell density was estimated by dividing the estimated cell number by the measured tissue volume.
Statistical analysis
For comparisons of normally distributed data between the two groups, we employed a two-tailed t test. To assess the main and interaction effects of postnatal age, sex, and IH exposure, we performed two-factor ANOVA. Post hoc Sidak's multiple-comparisons test was then used to identify specific group differences revealed by the ANOVA. Repeated-measures (RM) ANOVA was employed to analyze multiday behavioral data and the effects of drug injections within subjects. Statistical analyses were performed using MATLAB R2023b (MathWorks) and GraphPad Prism 10 (GraphPad Software). A significance level of p ≤ 0.05 was used.
Results
Hyperactivity and impaired motor learning in adult mice after neonatal IH
Adult female mice after neonatal IH were hyperactive with increased locomotion speed, mobile time, and traveled distance, whereas locomotion remained unaffected in male animals (Fig. 1A). A two-way ANOVA was performed to evaluate the effects of sex and IH on traveled distance. The results indicated a significant main effect for IH, F(1,33) = 25.33, p = 0.0025. The main effect of mouse sex and interaction between sex and IH were not significant. No significant effects were found for the time in the center zone, recorded as a measure of anxiety (Fig. 1B).
Tests for motor activity and motor learning. Increased locomotion in the open field test (A) was observed in females in the IH group but was not different in time spent in the open field central zone. B, Maximum speed in complex wheel (C) was similar in full wheel days 1–3 between the control and IH groups, but significantly changed in complex wheel days 4–7. The increase in the maximum speed between Days 4 and 7 was less in the IH group. D–F, Rotarod test. Latency to fall increased with test days was but not different between the groups. *p < 0.05, ANOVA post hoc test.
Motor learning was assessed with a complex wheel test. When a full wheel was presented to individual mice for 3 consecutive days, the maximum running speed increased and reached the plateau by Day 3 in both the control and IH groups (Fig. 1C). In the complex wheel with missing runs on Day 4, the maximum running speed decreased to a larger degree in the control group but was larger than in the IH group by Day 7. Intriguingly, IH mice displayed faster running speeds in the complex wheel on Day 4. However, video recordings revealed more missteps in this group, suggesting potential motor incoordination despite the increased speed. The interaction term between IH and testing day on maximum running speed was significant on RM ANOVA (F(3,72) = 4.067, p = 0.01). A two-way ANOVA was performed to evaluate the effects of sex and IH on the change in the maximum running speed between Days 4 and 7, with both sex (F(1,26) = 4.269, p = 0.048) and IH (F(1,26) = 0.3821, p = 0.0008) factors being significant. Post hoc analysis revealed a larger increase in maximum running speed in controls (Fig. 1D), indicating impaired motor learning mice after neonatal IH.
Accelerating rotarod, performed for 6 consecutive days, revealed successful motor learning of this task, indicated in increased time to fall (test time main effect, F(1.928,44.34) = 40.16, p < 0.0001) but was not different between control and IH groups for either males or females (Fig. 1E,F).
Decreased inhibitory and increased excitatory drive on Betz cells in the motor cortex of adult mice after neonatal IH
We performed whole-cell patch-clamp recordings in pyramidal Betz cells of the primary motor cortex in of adult mice after neonatal IH. The glutamatergic miniature excitatory postsynaptic currents (mEPSCs) were recorded at a holding potential of −70 mV (Fig. 2A,B). GABAergic mIPSCs were recorded from the same cells at 0 mV (Fig. 2C,D). The amplitude of mEPSCs was higher in the IH group compared with control (t = 8.001, df = 26, p < 0.0001), but the mEPSC frequency was not different between the groups (Fig. 2E,F). For IPSCs (Fig. 2G,H), the amplitude was not different between the control group and IH groups, but the IPSC's frequency was lower in the IH group (t = 3.202, df = 16, p = 0.0056).
Whole-cell patch-clamp recordings in primary motor cortex pyramidal neurons in adult mice after neonatal IH. A, B, examples of mEPSCs recorded at a holding potential −70 mV in control (A) and IH (B) groups. C, D, examples of GABAergic IPSCs recorded at a holding potential of 0 mV. EPSC amplitude (E), but not EPSC frequency (F), was increased in the IH group. IPSC amplitude (G) was unchanged, but the frequency (H) was decreased in the IH group **p < 0.01, ****p < 0.001, t test.
Age-dependent effect of GABA inhibition on glutamatergic transmission in motor cortex
The sequence of fEPSP’s responses, consisting of the fiber volley (FV) followed by the first short-latency fEPSP and often by the longer latency second fEPSP at the stimulation intensity over 500 µA, was reproducible in all subgroups and observed in both control aCSF and after picrotoxin application (Fig. 3A). First and second field potentials, but not FV, were completely blocked by 50 µM CNQX and, therefore, reflected monosynaptic (latency 4–6 ms) and disynaptic activation (latency >10 ms), respectively. There were no other components in field potentials detected, such as NMDAR, kainate receptor, or GABA receptor-mediated synaptic components after CNQX blocked AMPARs, and, therefore, these fEPSPs were pure AMPAR fEPSPs (similar to Wallace et al., 2014). FVs were completely suppressed in the presence of 1 µM of TTX. To ensure that the FVs were not age dependent or sensitive to picrotoxin, we compared FV amplitudes for all subgroups using single-way ANOVA (F(31,234) = 1.04, p = 0.42). This comparison did not reveal a significant influence of picrotoxin or age on FV amplitudes. FV amplitudes increased strictly linearly proportional to stimulation intensity in the input–output curves. Since both criteria were met, we concluded that a comparison of the absolute values of FEPSPs could be used for the subsequent analysis.
Effect of GABA inhibition on postsynaptic field potential in mouse motor cortex slices. A, Examples of fEPSP and response components origination at P8. A, Truncated stimulation artifact, B, fiber volley, C, fEPSP, D, second polysynaptic fEPSP. The legend shows inhibitors used to identify the origin of the components. B, Increase of monosynaptic fEPSP after picrotoxin in the IH group. C, Increase of DCP after picrotoxin in the IH group after P9. D, Membrane potential was increased after neonatal IH at P10–P13 and in adults but was similar to control values after picrotoxin application. *p < 0.05, two-way ANOVA post hoc.
To examine the developmental change of GABA receptor inhibition on glutamatergic transmission, we recorded fEPSPs in motor cortex slices before and after the addition of picrotoxin to perfusate between P7 and P13 when the developmental switch in GABA action in pyramidal neurons of cortical slices was expected (Peerboom and Wierenga, 2021). Picrotoxin slightly decreased fEPSP between P7 and P8 (Fig. 3B), suggesting excitatory action of GABA at this age. Starting at P9 in IH pups and P10 in controls, blockage of GABAARs by picrotoxin increased the amplitude of fEPSP (main effect of age on two-factor ANOVA, F(7,66) = 13.02, p < 0.0001), indicating a switch to inhibitory GABA action. The increase of fEPSP after picrotoxin was more than twice as large in the IH group (main effect of IH, F(1,66) = 9.700, p < 0.0027) and significant between P9 and P13 (post hoc p < 0.001), but not different in adults.
To investigate the effects of changes in tonic inhibition on glutamatergic transmission following IH, the resting DCP was monitored at the beginning of each sweep before the stimulation artifact. Once a stable DCP was achieved, picrotoxin was applied. Since there was no spontaneous activity at the 200 µM concentration of picrotoxin, the tonic change of DCP before stimulation after picrotoxin application could be attributed to the blockage of extrasynaptic GABAARs. The deflection of DCP from the baseline after 15 min of picrotoxin application switched from negative at P7 to positive after P10 (Fig. 3C) and was larger in the IH adult group than in control adults at 15 min after picrotoxin (ANOVA, F(5,53) = 6.253, p = 0.0001, post hoc p = 0.0067), consistent with the time of changing polarity of the response to picrotoxin, as it was observed for fEPSP.
Patch-clamp recordings in pyramidal neurons confirmed an increase in tonic inhibition mediated by extrasynaptic GABAARs by a larger decrease of resting membrane potential Vm in the IH group with the addition of picrotoxin to perfusate (Fig. 3D). The absolute values of Vm were larger (more negative) after IH both at P10P13 and in adults (RM ANOVA, F(3,54) = 10.17, p = 0.0001) than in age-matched controls. Vm values were reduced to those in the control groups at corresponding ages with the application of picrotoxin (no difference between post-picrotoxin groups on post hoc comparisons), indicating that the increased hyperpolarization after IH was due to the activity of extrasynaptic GABAARs. The effect was observed in neonatal animals after P9 and persisted with a further increase in adult animals.
Picrotoxin increased spontaneous MUA in IH but not in control mice in vivo
Following a low dose of 0.3 mM picrotoxin application on the cortex surface of the control group of mice, there was no significant change in MUA relative to the pre-picrotoxin baseline, 99.25 ± 22.5% (Fig. 4A; t test, ns). The mean increase after picrotoxin injection was 213.74 ± 26.15% (Fig. 4B; t test, p < 0.013) relative to the pre-picrotoxin baseline. Mean MUA change relative to baseline was higher in the IH groups than in the controls (Fig. 5C; t test, p < 0.009).
Examples and group data of extracellular multi-unit recording in the motor cortex of awake neonatal mice at P12–P14. Control (A) and intermittent hypoxia (B) animals are shown. After 9 min of baseline recording, 1 µl of 0.3 mM picrotoxin (PTX) was superfused on the exposed cortex above the recording electrode (arrow). C, MUA increase was significantly larger in the IH group. **p < 0.01, t test.
Changes in indexes of short-range functional connectivity after pharmacological GABA inhibition on rsfMRI at P12. A, Representative T2* images and corresponding local functional connectivity indexes before and after cortical picrotoxin injection in a mouse pup after IH. The arrow indicates an MRI-visible marker placed on the site of picrotoxin injection. Arrowhead points to the local increase of ReHo connectivity at the injection site. Oval ROI was placed on the site of injection. Scale bar, 1 mm. B–D, The individual animal changes in indexes of short-range functional connectivity after ALFF, fALFF, and ReHo in cortical regions of interest around picrotoxin injection. *p < 0.05, **p < 0.01, ***p < 0.01, RM ANOVA post hoc.
Increased ALFF and regional signal coherence in mouse pups motor cortex after neonatal IH
Resting-state fMRI was performed on awake mouse pups at P12. Representative maps of local functional connectivity indexes before and after cortical picrotoxin injection are shown in Figure 5A. A repeated-measures two-way ANOVA was performed to evaluate the effects of IH and pharmacological GABA inhibition after picrotoxin injection on indexes of local functional connectivity. The results for ALFF indicated significant main effects for IH, F(1,14) = 5.460, p = 0.034, and for picrotoxin F(1,14) = 17.36, p = 0.009. The interaction between IH and picrotoxin was not significant. ALFF was higher in the IH group (p = 0.04) before picrotoxin. Picrotoxin injection increased ALFF in the IH group (p = 0.0016), but not in the control group. Picrotoxin injection also increased fALFF (F(1,14) = 14.45, p = 0.0019, RM ANOVA, p = 0.005 post hoc) and ReHo (F(1,14) = 13.75, p = 0.0023, RM ANOVA, p = 0.0008 post hoc) in IH groups, but not in the control groups.
The pattern of changes in ALFF and ReHo suggested a potential for both local and global effects of cortical picrotoxin injection on ALFF and ReHo. While our analysis focused on the local region of interest around the injection site, to align with the primary focus of the study, there is a possibility of broader network-level effects. Due to technical limitations, such as the positioning of the surface coil and potential signal-to-noise ratio issues in peripheral regions, we did not attempt to quantify these global changes.
Increased gene expression in markers of glutamatergic transmission but not in markers of interneuron function
To elucidate the molecular mechanisms underlying the observed changes in interneuron-mediated regulation of glutamatergic transmission and neuronal synchronization, we compared gene expression profiles in the motor cortex of control and IH pups. We focused on a predefined set of genes involved in ion channel function and synaptic transmission. Our findings align with the previous reports (Goussakov et al., 2021) by demonstrating a significant upregulation of several AMPA receptor subunits in the IH group, consistent with the enhanced AMPA receptor-mediated transmission. Interestingly, we observed no significant changes in the expression of the GABA synthetic enzyme glutamate decarboxylase (GAD), including GAD1 and GAD2, which are considered reliable markers of GABA release (Dicken et al., 2015).
Similarly, no significant differences were observed between control and IH groups at key developmental stages in the expression of Slc12a2 and Slc12a5, genes encoding NKCC1 and KCC2 proteins that regulate the shift of GABA from excitatory to inhibitory. This finding aligns with our field potential recordings using picrotoxin, suggesting that IH does not affect the timing of the GABA switch. Interestingly, we observed a trend toward elevated expression of the delta subunit of GABA receptors in the IH group at P14 and P30, potentially linked to extrasynaptic tonic GABA inhibition, which aligns with our findings in the DCP and Vm at those ages (Fig. 3C). However, this increase did not reach statistical significance with ANOVA.
Long-term decrease in PV expression in the somatosensory cortex after neonatal IH
To examine changes in the number of inhibitory interneurons subtypes, we quantified the density of major interneuron types expressing parvalbumin (PV+) and somatostatin (SST+) immunoreactivity in the somatosensory cortex of mice at P12, 5 d after exposure to neonatal intermittent hypoxia (IH) between P3 and P7, and at P60 when the brain is mature. There was a significant decrease in the density of PV+ interneurons in the IH group compared with normoxic controls at P12 and at P60 (main effects for IH, F(1,8) = 67.44, p < 0.0001, post hoc p = 0.0045 for P12 and p = 0.0008 for P40; Fig. 6A). The density of neurons expressing PV increased with age both in control and IH groups (F(1,8) = 21.17, p = 0.0018). The interaction between IH and age was not significant. The difference between the control and IH groups in SST+ neuron density was not significant for both ages (F(1,8) = 1.601, ns; Fig. 6B). There was a decrease in the density of SST+ neurons in the P60 animals (F(1,8) = 8.504, p = 0.0194) relative to P12 in both control and IH groups.
Stereological count of interneurons, labeled for parvalbumin (A) and somatostatin (B) in the sensory-motor cortex at P13 after neonatal IH. **p < 0.01, ***p < 0.0-1, ANOVA post hoc.
Discussion
We investigated the interaction between major neurotransmission systems, GABA and glutamatergic, during a critical stage of normal cortical network development and after perinatal IH brain injury. The major findings of the study were (1) increased glutamatergic transmission after IH, evidenced by patch-clamp miniature currents and field potential recordings with a selective GABAARs antagonist picrotoxin; (2) decreased inhibitory drive after IH, based on lower IPSC frequency on pyramidal cells and MUA recordings with picrotoxin injection, immediate and long-term decrease of PV+ interneuron density; and (3) increased tonic inhibition by GABAARs in the motor cortex in the first week after neonatal IH and in adult age. The other significant result in this study was abnormally increased neuronal connectivity in the actively developing cortex at the second postnatal week after IH, reflected by the elevated low-frequency fluctuations in rsfMRI and further enhanced by picrotoxin. Long-term behavioral consequences, locomotor hyperactivity, and impaired motor learning were observed in adult mice exposed to neonatal IH.
Perinatal hypoxic brain injury occurs at the critical period of cortex development and activity-dependent formation of functional neuronal networks (Blankenship and Feller, 2010), governed by maturing excitatory and inhibitory signaling. We have previously reported abnormally increased excitatory glutamatergic AMPA- and NMDA-mediated transmission in the hippocampus after perinatal IH (Goussakov et al., 2021) and now confirmed it in the motor cortex by the increased miniature EPSC’s amplitudes in principal excitatory neurons. Maturation of the inhibitory GABAergic system occurs later in the development and the transition from excitatory to inhibitory action of GABA constitutes a major change in the pattern of neuronal activity (Ben-Ari et al., 2007). While there is strong evidence supporting the excitatory role of GABA during early postnatal development (Murata and Colonnese, 2020), the extent to which this occurs in vivo and its precise physiological significance remain areas of active research (Valeeva et al., 2016).
In our field potential recording with picrotoxin during the second postnatal weeks in mice, GABA transmission provided an excitatory effect before switching to inhibition between P9 and P10 both IH and control. The small difference between the groups did not support the notion that the timing of the GABA switch is influenced by neuronal activity (Fiumelli et al., 2005) and may be affected by neonatal IH. The relative gene expression of NKCC1 and KCC2 chloride transporters, mediating GABA switch, also did not differ between the groups.
After applying picrotoxin after P9, the magnitude of the monosynaptic fEPSP was nearly double in the IH group relative to controls. This inhibition affected the activation of neurons, measured in the short-latency monosynaptic response that does not involve either feedback or feedforward inhibitory circuits. The larger fEPSP changes likely result from two contributing factors: the previously described elevated AMPA transmission and a larger, picrotoxin-sensitive tonic GABAergic inhibition. This inhibition does not involve a phasic component, mediated by the synaptic release of GABA from interneurons, and it is most likely mediated by extrasynaptic GABA receptors (Marchionni et al., 2007). Further confirmation of the increased tonic GABAAR inhibition after neonatal IH was obtained by recording DCP and Vm before and after picrotoxin application. Consistent with the age-related changes in fEPSP, the deflection of DCP from negative to positive occurred between P9 and P10 in both P10–P13 pups and adults. The increased transmembrane potential due to the activity of extrasynaptic GABAARs was observed starting a few days after IH and persisted in adults. The upregulation of extrasynaptic GABAARs and the subsequent increase in tonic inhibition may represent a compensatory mechanism to counterbalance the heightened glutamatergic activity and AMPA receptor upregulation, consistently observed after neonatal hypoxia. This compensatory response could help to prevent excessive neuronal excitation and potential neurotoxicity. However, chronic elevation of extrasynaptic GABA can have deleterious consequences for neuronal function and plasticity.
GABAergic interneurons are the major population of cortical inhibitory neurons, representing ∼20% of all cortical neurons (Sahara et al., 2012). The effect of perinatal hypoxia on interneuron development remains controversial. An increase in the density of the perineuronal nets and PV+ interneurons was observed in adult rats treated with moderate hypoxia on the first day of life (Trnski et al., 2022). On the other hand, the numbers of cortical interneurons expressing immunohistochemically detectable levels of PV and SST and vasoactive intestinal peptide were decreased by P15 in hypoxic-reared mice (Komitova et al., 2013). Since the number of GAD+ neurons did not change, the authors suggested a delay in maturation, but not death of PV+ interneurons after hypoxia. However, a loss of GAD+ and PV+ interneurons and disruption of perineuronal nets was found in the cerebral cortex following hypoxia-ischemia in near-term fetal sheep (Fowke et al., 2018), at P18 in mice hippocampus after neonatal hypoxia-ischemia (Chavez-Valdez et al., 2018), in inflammation-induced perinatal injury in mice and humans (Stolp et al., 2019). The timing and severity of hypoxic insult may affect the numbers and maturation of interneurons in the developing cortex. Our study revealed decreased inhibitory drive due to an immediate and long-term decrease of PV+ (but not SST+) interneuron density and functional interneuron deficiency by decreased IPSC frequency. Considering that the expression of PV and maturation of PV+ neuron firing properties depends on the excitatory synaptic drive (Miyamae et al., 2017; Warm et al., 2021), it is not clear why the decreased inhibitory drive coincides with the elevated glutamatergic transmission in the cortex after perinatal hypoxic brain injury and requires further investigation.
In the current study, we demonstrate for the first time the functional deficiency of cortical interneurons after perinatal brain injury by recording on nonanesthetized neonatal animals with a GABA blocker application. A low dose of picrotoxin did not affect MUA in controls while increasing MUA in IH animals, suggesting a reduced capacity of the GABAergic inhibitory system to control excessive excitation and prevent hyper-synchronization of neuronal activity in the IH group. The observed functional deficiency in the inhibitory system and the increased glutamatergic transmission likely disrupt neuronal network formation, contributing to hyperactivity and motor learning deficits observed in this study and in previous reports after perinatal hypoxia (Mikati et al., 2005; Goussakov et al., 2021; Trnski et al., 2022).
Our finding of coexisting increased tonic inhibition with decreased synaptic inhibition highlights the complex regulation of the excitation/inhibition balance in the developing brain after injury. This interplay involves phasic (synaptic) and tonic (extrasynaptic) GABAergic modulation of glutamatergic transmission. Interestingly, despite the decrease in extrasynaptic inhibition, the overall effect of neonatal hypoxia on neuronal circuitry appears to be excitatory, as shown in the current and previous data (Jensen et al., 1998; Goussakov et al., 2021). A critical implication is that a substantial rise in excitatory transmission alone can lead to dysfunctional networks prone to seizures. Therefore, compensatory mechanisms, such as the observed robust elevation of tonic inhibition directly targeting monosynaptic and polysynaptic glutamatergic inputs, become essential for maintaining network functionality even during heightened excitatory drive. Compensatory changes in excitatory and inhibitory balance in the cerebral cortex has been described in various disease conditions but the exact mechanism of compensation depends on underlying pathology (Antoine et al., 2019).
We aim to assess the potential of resting-state fMRI for noninvasively monitoring somatosensory cortical network development after perinatal hypoxia. Local functional connectivity, measured by rsfMRI, likely depends on the activity and architecture of a small-scale network of primary cells, GABAergic interneurons of various types, astrocytes, and microvessels, where the BOLD signal originates. However, the mechanisms of those interactions and the contribution of specific cell types remain poorly understood. Short-range functional rsfMRI connectivity has been described to increase in fetuses and neonates (Cao et al., 2017; Huang et al., 2020) and then decrease later into adulthood (Ouyang et al., 2017), depending on the anatomical region (He and Parikh, 2016). The inverted U-shaped trend is often attributed to a combination of synaptogenesis followed by synaptic pruning, although the exact mechanisms remain unclear. Atypical development of local rsfMRI connectivity is associated with major cognitive and behavioral disorders, including ADHD (Sato et al., 2012; Sripada et al., 2014), schizophrenia, and autism (Fair et al., 2009).
The increased ALFF in the IH group likely reflects enhanced neuronal activity or synchronization. Ex vivo data suggests increased glutamatergic transmission may contribute to the former, while impaired interneuron function may contribute to the latter. Local GABA inhibition with picrotoxin further supports the role of interneuron dysfunction in synchronization, as it led to increased ALFF and ReHo in the IH group but not in controls at this dose of picrotoxin. The lack of a significant ReHo increase in the IH group without picrotoxin suggests that the inhibitory network may retain sufficient capacity for desynchronization during the resting state. A higher regional ALFF was found in infants born prematurely (Feng et al., 2024) and may be associated with higher levels of whole-brain connectivity, as observed in children with autism (Supekar et al., 2013) and attention deficit hyperactivity disorder (Yu-Feng et al., 2007). Neonates who have a first-degree relative with autism spectrum disorders had significantly higher ReHo within multiple resting-state networks in comparison with age-matched controls, preceding differences in long-range connectivity (Ciarrusta et al., 2020). Understanding the relationship between rsfMRI measures and functional cortical development will be critical for establishing this technique as a valuable tool for evaluating therapeutic interventions in neonates with hypoxic brain injury. This relationship can be complex considering that preterm birth has a highly heterogeneous impact on the developing brain (Dimitrova et al., 2020).
The study has several limitations. The electrophysiological studies were conducted on male-only mice to reduce the measurement variability due to the sex factor and achieve more statistical power in low-throughput electrophysiological experiments. Male mice were chosen as clinical reports increased the susceptibility of boys to neonatal hypoxia-ischemia, although revealed in a very large cohort (Chalak et al., 2023). The rest of the outcome measures were reported separately for males and females. Second, the study utilized mIPSC recording of spontaneous inhibitory synaptic activity. Adding evoked IPSCs to the study would provide valuable complementary information to assess the overall inhibitory synaptic strength and coordination, leading to a more comprehensive understanding of inhibitory synaptic function.
Overall, this study demonstrates that perinatal brain injury disrupts the balance of excitation and inhibition in developing cortical networks. This disruption, potentially caused by functional deficiencies in GABAergic interneurons alongside increased glutamatergic transmission, may contribute to altered brain connectivity and the observed behavioral deficits, including hyperactivity and cognitive difficulties. This study underscores the potential of rsfMRI as a valuable clinical tool to elucidate mechanisms of functional maturation and abnormal connectivity in the developing cortex following perinatal hypoxic brain injury.
Footnotes
This study was funded by National Institutes of Health grants 1R01NS119251-01A1, R01 NS107383, R01 GM112715, and S10OD032223.
The authors declare no competing financial interests.
- Correspondence should be addressed to Alexander Drobyshevsky at adrobyshevsky{at}northshore.org.
