Abstract
Neonatal hypoxia (Hx) causes white matter (WM) injury, particularly in the cerebellum. We previously demonstrated that Hx-induced reduction of cerebellar Purkinje cell (PC) activity results in locomotor deficits. Yet, the mechanism of Hx-induced cerebellar WM injury and associated locomotor abnormalities remains undetermined. Here, we show that the cerebellar WM injury and linked locomotor deficits are driven by PC activity and are reversed when PC activity is restored. Using optogenetics and multielectrode array recordings, we manipulated PC activity and captured the resulting cellular responses in WM oligodendrocyte precursor cells and GABAergic interneurons. To emulate the effects of Hx, we used light-activated halorhodopsin targeted specifically to the PC layer of normal mice. Suppression of PC firing activity at P13 and P21 phenocopied the locomotor deficits observed in Hx. Moreover, histopathologic analysis of the developing cerebellar WM following PC inhibition (P21) revealed a corresponding reduction in oligodendrocyte maturation and myelination, akin to our findings in Hx mice. Conversely, PC stimulation restored PC activity, promoted oligodendrocyte maturation, and enhanced myelination, resulting in reversed Hx-induced locomotor deficits. Our findings highlight the crucial role of PC activity in cerebellar WM development and locomotor performance following neonatal injury.
Significance Statement
Adult survivors of prematurity often experience locomotor incoordination secondary to cerebellar dysfunction. The cerebellum develops in the last trimester of pregnancy, a period that preterm neonates miss. Here, we show how neonatal hypoxia alters the crosstalk between neurons and oligodendrocytes in the developing cerebellum. Through loss-of-function and gain-of-function experiments, we unveiled that neuronal activity drives cerebellum-associated white matter (WM) injury and locomotor dysfunction after hypoxia. Importantly, restoring neuronal activity using direct neurophysiological stimulation reversed the hypoxia-induced white matter injury and locomotor deficits. Early cerebellar neuronal stimulation could serve as a potential therapeutic intervention for locomotor dysfunction in neonates.
Introduction
Preterm birth is frequently associated with both neuronal and white matter (WM) injury, possessing particular risk to the rapidly developing cerebellum. The cerebellum undergoes a critical phase of WM maturation during late gestation (Volpe, 2009). Moreover, preterm infants are exposed to long periods of Hx in the neonatal intensive care unit (NICU) due to their lung immaturity, which has been implicated in reduced and delayed myelination, and results in long-term cognitive impairments in this population (Yaari et al., 2012; Ferrarelli, 2015; Nakata et al., 2017; Marinelli et al., 2018; Sathyanesan et al., 2018). While cerebellar WM defects are well-characterized in animal models of preterm birth, the precise cellular and physiological mechanisms driving these developmental abnormalities remain largely unknown, although modulation of WM development is certainly implicated.
In humans, fetal oligodendrocyte precursor cells (OPCs) proliferate in the cerebellum during the third trimester (weeks 28–40). In the neonatal period (postnatal weeks 0–4), newly born OPCs differentiate into mature oligodendrocytes (OLs). This differentiation is crucial as it initiates axonal myelination, in which the OL processes extend around axons, forming myelin sheaths. The timing and extent of the myelination are intricately regulated by several neurotransmitters, including glutamate, GABA, and acetylcholine (Gallo and Deneen, 2014). The maturation of glial cells is likewise regulated by the aforementioned neurotransmitters, which evoke responses in WM OPCs via postsynaptic ionotropic and metabotropic receptors (S. C. Lin and Bergles, 2004; Butt et al., 2014) and neuron–OPC synapses, resulting in stage-specific white matter development (Kukley et al., 2010).
Notably, the activity of Purkinje cells (PCs) is integral in guiding white matter development, particularly in the neonatal period (postnatal weeks 0–4). PC axons represent the only output from the cerebellar cortex, and the activity of these cells influences neurotransmitter release, critical in both the extracellular signaling guiding myelination and in the modulation of glial cell maturation (Pajevic et al., 2014; Barron et al., 2018). We previously demonstrated that neonatal Hx induces functional impairment in cerebellar PCs and linked abnormalities in PC physiology with specific aspects of locomotor dysfunction (Sathyanesan et al., 2018, 2021). However, how PC activity relates to WM injury and locomotor function remains unclear. Interestingly, we have previously demonstrated that in the cerebellar WM, GABA plays a regulatory role in OPC development and myelination through transient synapses formed between migrating immature inhibitory interneurons and OPCs (Zonouzi et al., 2015). This modulation is disrupted by chronic neonatal Hx, which attenuates GABAergic signaling between migrating interneurons and OPCs in the cerebellar WM, leading to maturational deficits and hypomyelination (Zonouzi et al., 2015). Consistent with the crucial role of GABA in developmental myelination in the cerebellum, we have also shown that GABAergic drugs partially alleviate Hx-induced motor learning deficits and limb coordination anomalies (Sathyanesan et al., 2018). However, the mechanism by which PC axonal activity modulates the developing WM cells and ultimately locomotor performance and plasticity remain elusive. Previous work by McKenzie et al. (2014) underscored the crucial role of active myelination in the corpus callosum for motor skill learning in mice engaged in complex wheel performance.
In the present study, we sought to investigate a causal relationship linking PC activity to myelination and modulation of locomotor coordination, as well as plasticity. We hypothesized that synaptic axon–OPC interactions may play distinct roles in neurotransmitter-mediated modulation of OPC development and myelination in the postnatal cerebellar WM. By leveraging in vivo extracellular multiunit recording, we explored cerebellar WM OPC and GABAergic cell responses to PC activity during postnatal development—a critical period for locomotor learning (Birey et al., 2015; Sakry et al., 2015). Using optogenetics, we manipulated neuronal activity in cortical cerebellar cells and performed behavioral testing to assess the resulting impact on locomotor function, as well as histological analysis of OL development and myelination. Additionally, we examined whether suppressing PC axonal activity in normal, uninjured mice mirrored the cellular physiological changes and locomotor performance deficits observed in Hx-exposed mice. Lastly, we investigated whether direct stimulation and restoration of PC activity reversed the Hx-induced abnormalities in cerebellar OL development and associated motor defects.
Materials and Methods
Sex as a biological variable
This study utilized both male and female mice but sex was not considered as a biological variable.
Animals
Male and female Pcp2-cre mice [B6.129-Tg(Pcp2-cre)2Mpin/J; The Jackson Laboratory] were used for all in vivo optogenetics, multielectrode array extracellular recordings, and behavioral experiments. GAD2-cre [B6N.Cg-Gad2tm2(cre)Zjh/J] and PDGFRα-cre [B6.129S-Pdgfratm1.1(cre/ERT2)Blh/J] mice were used for identification and verification of GABAergic interneuron and OPC electrical activity, respectively. All animals were handled in accordance with the Institutional Animal Care and Use Committee (#00030473) of Children's National Medical Center, which approved all the protocols and the Guide for the Care and Use of Laboratory Animals (National Institutes of Health).
Hypoxic rearing
To induce neonatal brain injury, we reared all our mice for chronic hypoxia experiments in a sublethal hypoxia chamber. P2 mice pups were cross-fostered with CD-1 mothers and placed in a hypoxia chamber (10.5% fiO2) from P3 to P7 (unless mentioned otherwise). After hypoxia (Hx), mice were returned to standard laboratory housing conditions following a brief period of reacclimatization to normoxic (Nx) conditions.
In vivo extracellular electrophysiology combined with optogenetic technique
We used an in vivo multielectrode array for extracellular electrophysiology to record the activity of white matter (WM) cells. We followed our previous publication's protocol to optimize the adeno-associated virus (AAV) infection rate in GABAergic interneurons in the white matter of GAD2-cre mice (Sathyanesan et al., 2018). We calculated the viral expression count over the total number of GABAergic cells present within the white matter boundary of the field of view (NeuN+ cells). We repeated this analysis in several microscopic fields in the same mouse and repeated the analysis in five separate mice. Our results showed a 70% viral infection rate in white matter. We also analyzed viral expression in infected cells of Hx mice and increased the AAV-injected MOI (vg/ml) under these experimental conditions in order to compensate for the change in PC physiology after Hx (Extended Data Fig. 2-1A–E). We used the same AAV-injected MOI (vg/ml) in PDGFR-α mice, but we were unable to optimize the process in these mice because of their very low birth/survival rate after Hx. In order to study the responses of WM cells to Purkinje cell (PC) activity, we selectively expressed cre-depended channelrhodopsin tagged with mCherry fluorophore [pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE] from UPenn Vector Core (catalog #AV-9-20297P), injected stereotactically at A/P, −5.6 mm; M/L, ±2.1 mm; and DV, −0.35 mm into Nx and Hx Pcp2-cre mice bilaterally at P9 (unless mentioned otherwise). The channelrhodopsin expression pattern for both Nx and Hx mice is shown in Figure 1A. For GABAergic interneurons and OPC electrical activity identification and validation, the abovementioned virus was injected into the cerebellar white matter (A/P, −5.6 mm; M/L, ±2.2 mm; DV, −1.5 mm) of GAD2-cre and PDGFRα-cre mice at P9. For the experiments in which we suppressed PC activity, the PCs of Nx Pcp2-cre mice were transfected with pAAV-double floxed-eNpHR-EYFP-WPRE-pA (Addgene 20949), halorhodopsin at P9 (Fig. 1B).
Selective optical stimulation of cerebellar PCs results in the modulation of cellular activity in WM. A, B, Schematic representation of recording electrodes and optical fiber for optical stimulation in the mouse brain, made in Biorender. Before starting the in vivo extracellular electrophysiological recording in WM, selective stimulation of PCs was confirmed by placing an eight-electrode array in the PC layer. The representative raw traces depict the increase in PC spiking activity observed during the optical stimulation (t = 1 s; 25 Hz) with blue light (λ = 470 nm) and reduced spiking activity during stimulation (t = 2 s; continuous) with orange light (λ = 590 nm). C, The pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE virus and (D) pAAV-double floxed-eNpHR-EYFP-WPRE-pA virus injected locally into the cerebellar PC layer of Pcp2-cre Nx mice. The mCherry (C, in red) tagged with ChR2 and halorhodopsin tagged to YFP (D, in green) are selectively expressed in PCs stained with DAPI (blue). Scale bar, 50 µm. E, F, PC stimulation alters WM cellular activity in normal development at P13 and P21, respectively. At P13, Nx mice (n = 10) displayed an average basal firing frequency of 16.24 Hz, which was increased to 21.36 Hz an indication of excited cells, whereas in inhibited cells basal firing frequency was reduced to 16.68 Hz from 20.97 Hz. Excited and inhibited cells represented 47.66 and 13.28% of the total WM cells captured during extracellular recordings. In 39.06% of cells, no significant changes in firing frequency were observed during PC stimulation (nonresponsive cells). During the development of Nx mice (P13–P21), the percentage of excitatory cells decreased to 34.78%, whereas the percentage of inhibitory cells increased to 16.3%. G, H, Overall responses of WM cells, after PC stimulation, in chronic neonatal Hx in both P13 and P21, respectively (n = 16). Excited cells were not identified at P13 or at P21. The percentage of inhibited cells increased from 18.78 to 37.84% from P13 to P21. The bar graphs represent the mean firing frequency with (±) SEM of overall WM cell responses before, during, and after selective optical stimulation with blue light (λ = 470 nm; 1 s; 25 Hz) on pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE–expressing PCs of Nx and Hx Pcp2-cre mice. The pie charts show the percentage of WM cells that were excited, inhibited, or nonresponsive upon PC stimulation. I, J, Difference in WM inhibitory cell response between the Nx and Hx groups in both P13 and P21 mice, respectively, after PC stimulation. There are higher firing frequency rates of inhibitory WM cells in both P13 and P21 of Nx mice than in Hx mice before, during, and after stimulation of PCs. The average firing frequency of PC activity plots over time at two different developmental timepoints, P13 and P21, and the alteration of PC spontaneous electrical activity after optogenetic stimulation are shown in Extended Data Figure 1-1A–F. The graphical illustration in Extended Data Figure 1-2 shows the proposed mechanism by which increasing firing of PC leads to reduced inhibition of some WM cells, which results in the excitation of nearly 50% of cells in the WM of Nx mice.
Figure 1-1
Cerebellar PC spontaneous electrical activity alter with chronic optogenetic stimulation. The average firing frequency of PCs activity plots over time at two different developmental timepoints, P13 and P21. n = 8-10 animals/group. (A) The spontaneous firing frequency of Nx (pink) PCs increases with blue light (λ=473 nm; 25 Hz, 30msec pulse duration, 25 pulses) selective stimulation of pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE expressed PCs at P13. Hypoxia injury reduces the PCs electrical activity (purple) almost three folds down for age matched mice, which does not evoke any additional increase in frequency during optical stimulation. (B) At P21 PCs exabits robust spontaneous firing frequency of 54.1 Hz and impressive evoked frequency of 77.36 Hz during blue light stimulation. Ten days post hypoxia injury with normal development does not recover the physiological deficits of PCs (purple), in terms of both the basal and evoked activity. (C) and (D) show the average firing frequency of PCs decrease during selective orange (λ=590 nm; continuous exposure for 1000msec) light exposure to pAAV-double-floxed-eNpHR-EYFP-WPRE-pA transfected Purkinje cells at P13 and P21, respectively. The down regulation of PCs activities is very coinciding with the duration of orange light exposure. Unlike PCs activation (pink) during blue optical stimulation, indicates that, once the cells start firing fast, the membrane threshold potential goes up and require some addition time to bring the membrane potential down to resting stage. This physiology reflects in optogenetic activation (A, B), where the firing frequency does not reach to the spontaneous label immediately at the end of the blue light stimulation. (E, F) Bar graphs of the PC firing frequency during blue and orange light pre-stim, stim, and post-stim of Nx and Hx mice at P13 and P21, respectively. The period of optical exposure in respective graphs occurred from 0-1000 msec. Download Figure 1-1, TIF file.
Figure 1-2
Graphical Illustration of PC interaction with WM interneurons and OPCs. Activation of PCs results in the inhibition of inhibitory GABAergic interneurons, which can ultimately produce excitatory postsynaptic current via disinhibition in OPCs receiving interneuron input, as opposed to OPCs receiving direct input from PCs that would be inhibited by the GABA molecules released from PCs. In this model, increased firing of PCs leads to reduced inhibition of some WM cells, resulting in observed excitation of approximately 50% of cells in the WM in Nx mice. Download Figure 1-2, TIF file.
Figure 1-3
Schematic representation of PC stimulation and inhibition effect on GABAergic neurons and OPCs. This alternative schematic illustrates that activation of PCs results in the inhibition of inhibitory GABAergic interneurons in both Nx and Hx Stimulated mice. As a results, this produces excitatory postsynaptic current through disinhibition in OPCs that are receiving interneuron input and ultimately leads to increased OPC maturation. Download Figure 1-3, TIF file.
For in vivo extracellular recordings, we used 1 MΩ tungsten electrode of eight (Microprobes MEA) coupled with a 200-µm-diameter optical fiber, connected to either a 473 or 590 nm wavelength LED light (PlexBright LED 4 channel optogenetic controller with radiant software, version 2.0, Plexon). One of the eight-electrode arrays along with the optical fiber was stereotactically inserted into the cerebellar PC layer (A/P, −5.6 mm; M/L, ±2.1 mm; DV, −0.35 mm) of isoflurane-anesthetized, head-fixed, virus-injected Pcp2-cre mice to selectively stimulate (λ = 473 nm LED) or suppress (λ = 590 nm LED) PC activity and simultaneously record the spontaneous and optically evoked activity. For all electrophysiological recordings (except for experiments in which PC activity was suppressed), we used 25 pulses of 30 ms duration each with a 25 Hz pulse frequency of blue light, as previously optimized for PcP2-cre mice PCs17. To capture the responses of these optogenetically modulated PCs in WM, we inserted another eight-electrode array into the WM, at the base of the same lobule (A/P, −5.6 mm; M/L, ±2.2 mm; DV, −1.5 mm) where we inserted the fiber optics for PC stimulation or suppression (Fig. 1C,D). All the electrophysiological data were acquired by OmniPlex D Neural Data Acquisition System as well as the original Multichannel Acquisition Processor Data Acquisition System control software from Plexon.
The electrophysiological recording was then postprocessed offline. The spike sorting was done by Offline Sorter v3.3 (Plexon) using peak-valley and template matching techniques. The waveform templates were created based on principal component analysis (PCA) and spike shape match. The frequency (spikes/s) was calculated in NeuroExplorer (Plexon) and plotted over time in Origin 9.0 graphics software. The duration of LED light exposure (blue or orange) for optogenetics application was denoted as a rectangular box of blue or orange, in respective figures.
Selective suppression and stimulation of Purkinje cell activity
For selective suppression of PC activity, we transfected healthy Nx Pcp2-cre mice with pAAV-double floxed-eNpHR-EYFP-WPRE-pA (Addgene 20949) at P9. The implantable ceramic (zirconia) cannulas with 2 mm fiber optics (purchased from Thorlabs, CFMXA05) of 200 µm core and 0.22 NA were then cut flat and polished to 0.35–0.4 mm in length using fiber termination and polishing kit (Thorlabs). After completion of virus injection, two cannulas with customized lengths were implanted stereotactically in the PC layer (A/P, −5.6 mm; M/L, ±2.1 mm; DV, −0.35 mm), one in each hemisphere and secured in place with dental cement. The AAV was incubated for 4 d to get the highest levels (60%) of expression of opsin on targeted cells, as described in our previous study17. At the age of P13, the PC activity suppression protocol was started by attaching the fiber patch cord with 200 µm core, 0.22 NA, 1×2 multimode couplers (Thorlabs, TM200SS1B) through a rotary joint to expose both hemispheres at once. The ceramic ferrule end was then connected to the implanted cannula on P14 Nx mice and secured with ceramic sleeves (Thorlabs). The other end of the patch cord with FC/PC connector was then attached to the orange (λ = 590 nm) LED light (PlexBright LED 4 channel optogenetic controller with radiant software, version 2.0, Plexon). The orange light exposure was continuously on for 60 s and kept off for 30 s to avoid heating. The paradigm was continued for 30 min for each hemisphere, once every day and consecutively for 7 d. At P21, the mice were either subjected to in vivo electrophysical recording or behavioral testing, followed by perfusion for immunohistochemical analysis (see below). To perform an in vivo electrophysiological analysis of P13 Nx mice in which PC activity was suppressed, we modified the protocol to fit into the experimental time frame. Nx Pcp2-cre mice were injected with channelrhodopsin at the age of P6–P7, and then the PC activity suppression protocol was started at P10 for 5 consecutive days following the same paradigm described above. At P14, extracellular recordings were performed 6 h after the end of the last PC activity suppression protocol.
For selective stimulation of PCs, Hx Pcp2-cre mice (P3–P7) were stereotactically injected with pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE (UPenn Vector Core, catalog #AV-9-20297P) into the PC layer (A/P, −5.6 mm; M/L, ±2.1 mm; DV, −0.35 mm). Two cannulas on both hemispheres were implanted as described above. Instead of orange light, in this set of experiments, a blue (λ = 473 nm) LED light (PlexBright LED 4 channel optogenetic controller with radiant software, version 2.0, Plexon) was used with 25 Hz, 30 ms pulse width for 120 s on and 30 s off to avoid local heating. The paradigm was then continued for 30 min for both hemispheres once a day for 7 consecutively days (P14–P20). At P21, the animals were tested for motor performance and then killed, and immunohistochemistry was performed.
Electron microscopy analysis
The brains of mice fixed with 4% PFA (n = 2–3) from all four groups (Nx, Hx, Hx stimulated, and Nx muted) were washed with PBS and then placed in 0.5% glutaraldehyde in 0.2 M cacodylate buffer overnight. Sagittal brain sections (300 μm) were obtained using a vibratome, washed in cold cacodylate buffer (5× 3 min) and then incubated in 3% potassium ferrocyanide in 0.1 M cacodylate buffer with 4 mM calcium chloride and 2% aqueous osmium tetroxide for 1 h at 4°C. This was followed by exposure to a filtered thiocarbohydrazide solution for 20 min at room temperature, with subsequent washing in ultrapure water (5× 3 min) and then incubation in an aqueous 2% osmium tetroxide solution for 30 min at room temperature. After the second osmium exposure, the sections were washed with ddH2O and then placed overnight in 1% uranyl acetate at 4°C. The next day, staining with Walton's lead aspartate solution was performed en bloc. The samples were dehydrated through a graded ethanol series (50, 70, 85, 95, and 100%), followed by placement in propylene oxide and infiltration with EPON resin (EMbed-812; Electron Microscopy Science), and then embedded between thermoplastic fluoropolymer films (Aclar, EMS). The sections were then placed in an oven at 60°C for 48 h for polymerization. Blocks containing the cerebellum lobules were trimmed and then sectioned using an ultramicrotome (z = 120 nm; UC7 Leica Microsystems). Ultrathin sections were placed on silicon wafers and carbon-taped in aluminum stubs for SEM imaging in a Helios NanoLab 660 FIBSEM (Thermo Fisher Scientific). To maximize the collection of backscattered electrons, a concentric detector in immersion mode was used at a 4 µm working distance, with 2 kV and 0.40 nA current landing. Tile-stitched images were generated using MAPS software (Thermo Fisher Scientific) at a pixel size of 1.6869 nm. For the myeline analysis, each electron microscopic image was zoomed in by 50%, in Fiji ImageJ processing software. The representative images from each experimental group were presented, and the defect in myelination was also marked with a red arrow. The parameters set for measurements were mean gray value, area, and perimeter in ImageJ. With the aid of the ROI manager tool, the inner and outer diameters of each axon/fiber were marked and measured from every field of view. Eight to 10 fields of view were measured to cover the whole acquired image. Over 450 axonal measurements from each experimental group were filed to calculate the g-ratio of cerebellar white matter and plot the bar graphs to compare between four experimental groups. The g-ratio was calculated by the ratio of the inner diameter of the fiber over the outer diameter of the same fiber. Nonparametric one-way ANOVA statistical analysis was performed between groups. A scatter plot of the g-ratio was also plotted against the inner diameter of each axonal fiber for all experimental groups. We also counted the number of unmyelinated axonal fibers over the total number of fibers that appeared in each field of view and represented this as “% of unmyelinated fiber” for each experimental group. Nonparametric one-way ANOVA was also performed to compare the groups.
Behavioral testing
All the mice followed either Nx or Hx protocol until P21 for their locomotor performance on the inclined beam test and rotarod test. In the inclined beam test, we followed the same protocol described previously (Carter et al., 2001; Brooks and Dunnett, 2009). A wooden beam of 80 cm in length was placed at an elevated 30° angle. Two different widths of wooden beams were used, i.e., 1 and 2 cm. A dark chamber with bedding and food was placed at the end of the incline as a target for the mice to reach. We recorded the behavioral performances of all Pcp2-cre mice (Nx, Hx, Nx with PC activity suppressed, and Hx with PC activity stimulated) and documented the number of foot slips—either hind legs or front legs—and the time to traverse the beam. The inclined beam test was performed at P21, as previously described (Scafidi et al., 2014). The age of P21 was chosen because this sensory–motor task is dependent on subcortical WM function and Nx mice or Hx mice cannot perform the task at a younger age.
We also performed a rotarod test with gradually accelerating speed using the rotarod apparatus (Panlab S.L.U.) to evaluate motor balance and coordination of all our experimental groups (Nx, Hx, Nx with PC activity suppressed, and Hx with PC activity stimulated) at the age of P21. After one trial of adaptation, the mice were placed on the rotating rod for three trials at a starting speed of 4 rpm, which was gradually increased to 40 rpm in 5 min. The time that an animal was able to hold itself on the rod with increased speed was recorded as the latency to fall (Brooks and Dunnett, 2009).
Immunohistochemistry
Immunohistochemical analysis was performed for all mice at P21. Mice were anesthetized with isoflurane and transcardially perfused with 0.1 M phosphate-buffered saline (PBS), pH 7.4, followed by 4% PFA. Brains were postfixed in 4% PFA overnight and coronally sectioned using sliding microtome at 40 µm thick and stored at 4⁰C in PBS until use.
Immunohistochemistry was performed on free-floating sections, which were blocked for 1 h in PBST (0.1% Triton X-100) and then into 20% normal goat serum (NGS), followed by overnight incubation at 4°C in primary antibodies diluted in PBS with Tween 20 (PBST) and 5% NGS. The primary antibodies we used were as follows: 1:250 dilution of rabbit anti-NG2 (Millipore) and 1:250 for mouse anti-CC1 (Calbiochem). The brain sections were incubated with species-appropriate secondary antibodies (1:500) for 2 h at room temperature. The secondary antibodies were Alexa Fluor 647-donkey anti-mouse (Jackson ImmunoResearch) and Alexa 594-donkey anti-rabbit (Jackson ImmunoResearch). The brain slices were then mounted with Vectashield mounting medium with DAPI (VectorLabs) on slides for microscopic analysis.
The myelin staining was performed by submerging the 40 µm thick coronal section of brain slices into the FluoroMyelin, green fluorescent myelin stains, 300× (Molecular Probes, Invitrogen; 300-fold dilution in PBS) for 2 h at room temperature.
All the immunohistochemistry images were taken under a BX61/63 automated upright fluorescence microscope (Olympus). Images were viewed using NIH ImageJ for postcounting. All histological quantifications were performed in an unbiased manner by investigators who were blinded to the experimental groups of the study. Immunolabeled cells were manually counted on the ImageJ “Cell Counter” plugin. In cerebellar white matter, six bilateral fields were analyzed from each brain section, and 3–5 brain sections/brain from 7 to 8 animals per group were randomly selected for this quantification. For each image, the total number of cells was counted and normalized with the total number of nuclei stained with DAPI. The green fluorescence intensity was estimated using ImageJ over the area for the FluoroMyelin staining.
Statistical analysis
Specific numbers of animals used for each experiment are denoted in the respective figure legend. Data were compiled and organized using Origin 9.0 software. All the graphs were plotted in GraphPad Prism as whole values with (±) SEM (standard error of the mean). Significance was calculated using the same software. The behavioral data and frequency plots were statistically compared using one-way ANOVA to determine whether overall differences existed across study groups of age-matched mice. An unpaired t test was conducted between pairs of mice groups’ and pairs of cell types’ spike durations as well as frequency expressions. Comparisons between groups were treated as unplanned comparisons and adjusted using Turkey's correction. The degree of statistical significance was denoted using asterisks (*p < 0.5; **p < 0.01; ***p < 0.001). The nonsignificant difference was denoted as n.s. We conducted four normality tests, D’Agostino and Pearson, Anderson–Darling, Shapiro–Wilk, and Kolmogorov–Smirnov tests. For sample sets that did not pass the normality tests, we conducted a nonparametric, Kruskal–Wallis, ANOVA, or Mann–Whitney t test. For those that did pass the normality tests, we conducted a parametric ANOVA or t test to ensure that our data did not deviate from normality and assume Gaussian distribution. ANOVA tests were conducted on groups of n = 3 or more while the t tests were conducted on groups of n = 2.
Results
PCs’ activity modulates cellular responses in the developing cerebellar WM
To investigate the influence of cerebellar PC activity on WM physiology, we devised a novel approach for in vivo extracellular electrophysiological recording of the WM during optogenetic manipulation of PCs. We used adeno-associated virus (AAV) vectors to express cre-dependent channelrhodopsin-mCherry (ChR2-mCherry) or halorhodopsin-EYFP (eNpHR-YFP) in Pcp2-cre mice, enabling either optogenetic stimulation or suppression of PC spiking activity, respectively (Extended Data Fig. 1-1A–F), as previously demonstrated (Sathyanesan et al., 2018). PC spiking activity was optogenetically manipulated while extracellular activity was recorded in WM using an in vivo multielectrode array placed at the base of the same cerebellar lobule (Fig. 1A–D).
Upon PC stimulation of P13 normoxic (Nx) mice (n = 10 mice, 128 cells) with ChR2-mCherry, we recorded three distinct types of WM responses. Most frequently, we observed evoked excitatory responses in WM spiking frequency (47.66%), with an increase from 16.24 to 21.36 Hz (one-way ANOVA; R2 = 0.4466; p < 0.0001). This may indicate that activation of PCs results in the inhibition of inhibitory GABAergic interneurons, which can ultimately produce excitatory postsynaptic current via disinhibition in OPCs receiving interneuron input, as opposed to OPCs receiving direct input from PCs that would be inhibited by the GABA molecules released from PCs (Extended Data Fig. 1-2). Less frequently, WM recordings demonstrated an inhibitory response to PC stimulation (13.28%), with a significant decrease in firing frequency from 20.97 to 16.68 Hz (one-way ANOVA; R2 = 0.3239; p < 0.0006). The remaining WM recordings (39.06%) showed no significant changes in firing frequency during PC stimulation and were classified as “nonresponsive” cells (Fig. 1E). However, at P21 (n = 10 mice, 92 cells), PC stimulation resulted in a lower proportion of excitatory WM response (34.78%, from 8.73 to 18.19 Hz, one-way ANOVA; R2 = 0.7093; p < 0.0001), a higher proportion of nonresponsive cells (49%), and a similar proportion of inhibitory WM responses (16.3%, 11.63 to 6.94 Hz; one-way ANOVA; R2 = 0.4961; p < 0.0001; Fig. 1F). Interestingly, a decrease in baseline peak frequency was observed with age; at P13, average frequency was 16.24 Hz, and at P21, this decreased to 8.83 Hz.
We next assessed the impact of chronic neonatal Hx on cerebellar signaling by recording WM spike frequencies. After PC stimulation, no evoked excitatory responses were observed in WM cells at either P13 or P21 (n = 16 mice, 671 and 301 cells, respectively; Fig. 1G,H). All cells either exhibited a nonresponsive pattern or demonstrated an inhibitory response. Furthermore, the percentage of inhibitory responses increased from 18.78% at P13 to 37.87% at P21 (Fig. 1G,H), indicating an increased inhibitory response induced by PC activity after neonatal Hx. Both P13 and P21 Nx mice exhibited a higher frequency signal of inhibitory WM cells after PC stimulation than in Hx mice (Fig. 1I,J), illustrating the impact of hypoxic conditions on cerebellar white matter cellular activity.
In summary, these results suggest that PC activity exerts a mix of excitatory and inhibitory effects on WM cell activity in a developmentally regulated manner, with the net excitatory influence diminishing with age. However, neonatal Hx induces a well-defined and persistent pathology in WM cell activity several weeks after the initial insult, and—as any excitatory influences are abolished—a net inhibitory response becomes more prominent.
Identification of cell types responsible for PC-induced neural activity in the developing cerebellar WM
We aimed to identify the specific cell types influencing WM activity under Nx and Hx conditions. Our previous findings demonstrated a regulatory role for immature GABAergic interneurons on OPCs. Developmentally regulated synaptic contacts between GABAergic interneurons and OPCs can regulate proper oligodendrocyte maturation and myelination in WM. In addition, both cell types are known to actively proliferate and migrate in the developing cerebellar WM, until the GABAergic interneurons reach the cerebellar cortex by P20 (Zonouzi et al., 2015). Based on these findings, we hypothesized that OPCs and GABAergic interneurons contribute to the differences in cellular electrical activity observed during cerebellar WM development (Zonouzi et al., 2015).
To test this hypothesis, we utilized an optogenetics-dependent approach to characterize the spike profile and action potentials (APs) of GABAergic interneurons and OPCs in cerebellar WM (S. C. Lin and Bergles, 2004; Bakiri et al., 2009). We electrophysiologically profiled and identified these cells based on validated tracings from previous studies (Guo et al., 2010; D. Fan et al., 2012; H. Fan et al., 2017). AAV vectors expressing ChR2-mCherry were utilized in GAD2-cre and PDGFRα-cre mice (Fig. 2A,B) to optogenetically stimulate GABAergic interneurons and OPCs in the cerebellar WM. We used an eight-electrode array with single optic fiber to record extracellular firing activity in WM. Subsequent offline analysis focused on cells displaying increased evoked activity during optical stimulation in each animal strain. Peak-valley spike sorting and principal component analysis (PCA) were applied to the recordings of spontaneous activities from the selected cells and revealed two distinct waveform profiles from GABAergic interneurons and OPCs in GAD2-cre and PDGFRα-cre mice, respectively (Fig. 2C,D).
Immunohistochemical and electrophysiological phenotype of WM cells responding to PC stimulation. Cells were identified as GABAergic interneurons OPCs, based on the expression of cell-specific markers, as well as spike waveform and PCA analysis of firing activity. A, B, Immunostaining of cerebellar WM cells demonstrating selective coexpression of ChR2-mCherry (red) in NeuN-expressing cells (green) in GAD2-cre mice and Olig2-expressing cells (green) in PDGFRα-cre mice. C, D, Extracellular responses (raw traces) recorded from WM cells of Pcp2-cre mice are influenced by selective PC stimulation. Responses were shortened, based on differences in waveform shape, amplitude, and duration. The spike profiles were compared with either pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE–expressing GABAergic interneurons in WM of GAD2-cre mice and pAAV-Efla-DIO-hChR2(H134R)-mCherry-WPRE–expressing OPCs in WM of PDGFRα-cre mice. Sorted spike waveforms from optically stimulated ChR2-mCherry–expressing GABAergic interneurons of GAD2-cre mice and OPCs of PDGFRα-cre mice (respectively) were compared with the responses recorded from Pcp2-cre mice in which cells were recorded upon PC stimulation. The representative plots of principal component 1 versus principal component 2 of the sorted waveforms from a single trace are shown for GABAergic interneurons from both GAD2-cre and Pcp2-cre mice and OPCs from PDGFRα-cre and Pcp2-cre mice, as well. The consistency of the waveform shape, quantified by direct comparisons of its principal components within pairs, confirmed that these were the same cellular types recorded under all conditions. Calibration, 500 mV × 0.1 ms. E, F, Measurements of the mean duration of waveforms with (±) SEM indicate no significant differences between GAD2-cre and Pcp2-cre mice and between PDGFRα-cre and Pcp2-cre mice. Calibration, 100 µm and 100 µs. Optimization of the viral infection rate was ensured through our previous protocol using an AAV infection rate as shown in Extended Data Figure 2-1A–E.
Figure 2-1
Optimization of viral infection rate, MOI (vg/ml) in cerebellar cells. (A, B) We ensured optimal viral infection rate of Purkinje cells (PCs) by following our previously published protocol using an AAV infection rate, MOI of 1 ul of 1X10^13 vg/ml stock and confirmed the expression rate of AAV in PCs in both Nx and Hx mice. Representative immunofluorescence staining using a viral infected Pcp2-cre cells (red)/calbindin (green)/ DAPI (blue). (C) Following the same protocol for oligodendrocyte progenitor cells, we confirmed optimization of viral infection rate, MOI (vg/ml) through immunofluorescence staining on viral infected PDGFRα-cre cells (red)/Olig2 (green)/DAPI (blue). Representative images of a Nx and a Hx murine cerebellum are shown here. (D, E) The same AAV dosage protocol was used for GABAergic interneurons in white matter of Nx GAD2-cre mice, where A. immunofluorescence staining of a viral infected GAD2-cre cells (red)/PAX2 (green)/DAPI (blue) and E. quantification of viral expression in infected interneurons in WM showed a 70% infection rate. We increased the AAV dose to 1.2 ul of 1X10^13 vg/ml stock under Hx conditions in an attempt to maintain optimal infection rates, but we were unable to further optimize the process in these mice, because these animals had very low birth and survival rates after Hx. Download Figure 2-1, TIF file.
These waveform profiles served as templates to discern the relative contributions of GABAergic interneurons and OPCs in PC-evoked WM responses recorded from Pcp2-cre mice. Out of several spike sets in WM recordings from Nx Pcp2-cre mice at P13, two were template-matched with the preidentified waveform template of GABAergic interneurons and OPCs. Cluster segregation analysis on principal component 1 against principal component 2 plots (Fig. 2C,D) allowed sorting of the waveforms and revealed one profile resembling the waveforms obtained from GAD2-cre mice (Fig. 2C) and a distinct profile resembling the waveforms from the PDGFRα-cre mice (Fig. 2D). OPCs exhibited no significant difference in spike waveform expression between PDGFRα-cre and Pcp2-cre mice, with a spike duration average of 771 and 583 µs, respectively (Fig. 2E). GABAergic interneurons showed a similar waveform duration between Pcp2-cre and GAD2-cre mice (Fig. 2F). These data are consistent with our hypothesis that GABAergic interneurons and OPCs are the primary contributors to the cellular activity in WM evoked by PC stimulation.
Functional contribution of GABAergic interneurons and OPCs to PC-induced cerebellar WM responses during normal development and after Hx
We next sought to investigate the impact of developmental age and Hx on the relative contributions of GABAergic interneurons and OPCs to PC-induced WM activity. Waveforms derived from Nx Pcp2-cre mice showed average durations, from onset of depolarization to completion of repolarization, of 400 and 818 µs for GABAergic interneurons (shown in red) and OPCs (shown in green), respectively (Fig. 3A,B). However, under Hx conditions, both GABAergic interneurons and OPCs showed shorter durations of action potentials, decreased to 243 and 429 µs, respectively (Fig. 3C).
Relative functional contribution of GABAergic interneurons and OPCs toward overall WM response to PC stimulation. A, The schematic represents the placement of the optical fiber (blue) for PC stimulation and the electrode array (red dots) on the mouse brain slice (not up to scale) for recording from WM at P13 and P21. B, C, Each multiunit extracellular recording from WM is sorted, based on the identification and validation methods for GABAergic interneurons and OPCs. The spikes are separated for both the Nx and Hx Pcp2-cre mice groups. The two representative profiles show the differences in spike shape and waveform duration of WM GABAergic interneurons (red) and OPCs (green). Bar graphs demonstrate that the duration of action potentials of GABAergic and OPCs of Nx mice were reduced in Hx conditions. D, E, Heat maps show the basal firing frequency distributions of all sorted waveforms, cell by cell over time (2,000 ms) for Nx and Hx mice, respectively. Strikingly the GABAergic interneuron average basal firing frequency was consistently between 20 and 60 Hz, whereas the OPC average basal firing frequency fell under 20 Hz. The pie charts indicate the percent of cells that display high firing frequency (light blue) and low firing frequency (dark blue). At P13, the functional contribution from GABAergic interneurons is 86.88 and 13.12% from OPCs for Nx mice. A switch is observed at P21, with 21.88% for GABAergic interneurons and 78.31% for OPCs. In Hx mice at P13, a decrease in the contribution of GABAergic neurons was observed (22.22%), with a concomitant increase in OPC contribution (77.78%). At P21, no GABAergic interneuron signature spikes were observed in WM recordings.
Further characterization of the spontaneous firing frequency of each WM cell revealed two frequency-dependent cell categories: high-frequency (20–60 Hz) and low-frequency cells (below 20 Hz). All GABAergic interneurons that met the criteria were identified as high-frequency cells, while OPCs exhibited low-frequency firing patterns. After PC stimulation at P13 in Nx (n = 10 mice, 60 cells), 86.88% of spike recordings originated from high-frequency GABAergic interneurons, while only 13.12% of spikes were initiated by OPCs. This ratio was inverted at P21 (n = 8 mice, 32 cells), where the most responsive cells were low-frequency OPCs, responsible for 78.13% of action potentials, and GABAergic interneurons contributed the remainder (21.88%; Fig. 3D). The predominance of OPC-derived action potentials is consistent with the observation that a lower frequency of PC-induced WM spikes was observed at P21 than at P13 (pre-stim average frequency was 8.73 Hz at P21 whereas 16.24 Hz at P13).
After neonatal Hx, the cellular distribution of WM responses was significantly altered during development. Although GABAergic interneuron signals predominated at P13 in the Nx group, in the Hx group at P13 (n = 12 mice, 126 cells), most cells responding to PC activity in WM were low-frequency OPCs (77.78%), and the remainder were GABAergic interneurons (22.22%; Fig. 3E). This early-onset OPC predominance persisted and was notably accentuated at P21, with all PC-responsive cells identified as OPCs (100%, n = 10 mice, 114 cells). Because OPCs demonstrate lower spike frequency than GABAergic interneurons, there is overall decreased PC-induced white matter activity following neonatal Hx than there is in Nx.
These findings align with prior work demonstrating the significant impact of Hx on GABAergic synaptic activity and the number of GABAergic interneurons in the developing cerebellar WM (Zonouzi et al., 2015). These spike-profiling data suggest that WM physiological activity corresponds to the relative contributions of GABAergic interneurons and OPCs.
Chronic reduction of PC electrical activity during development mimics Hx-induced electrophysiologic responses in WM
We recently discovered a persistent reduction in PC activity associated with Hx (Sathyanesan et al., 2021). Moreover, our findings offer compelling evidence that sustained suppression of PC activity can phenocopy Hx. To establish the significance of PC axonal inputs in WM development and physiology in vivo, we expressed cre-dependent halorhodopsin-EYFP and optogenetically suppressed PC activity for 5–7 d (Fig. 4A,B). The average waveform duration was 320 µs for GABAergic interneurons and 832 µs for OPCs (Fig. 4C). Chronic reduction of PC activity inhibited baseline WM responses at both P13 (n = 10 mice) and P21 (n = 8 mice; Fig. 4D).
Suppression of PC activity in normal Pcp2-cre mice reproduces the effects of Hx on WM cell response. A, The schematic indicates the placement of optic fiber (orange) in the mouse brain slice (not up to scale) to selectively inhibit PCs in pAAV-double floxed-eNpHR-EYFP-WPRE-pA mice by optical stimulation. The recording eight-electrode array placement in WM is shown by red dots. B, C, Representative spike profiles show two distinct waveform shapes, corresponding to GABAergic interneurons (red) and OPCs (green), with waveform templates with an average of 200 and 800 µs spike durations, respectively. The bar graph demonstrates lower waveform durations of GABAergic interneurons than OPCs under PC inhibition. D, The mean firing frequency graphs over time indicate inhibitory responses recorded in WM cells throughout the duration of inhibition of PC activity (0–10,000 ms) in both P13 and P21. Bar graphs of the firing frequency in both P13 and P21 before and during inhibition of PCs. E, The firing frequency distribution heat maps indicate that, at P13, both GABAergic interneurons (29.82%) and OPCs (70.18%) contributed toward WM cell response, but at P21, all the inhibitory responses were from OPCs alone. Scale bar, 100 µs.
Throughout the period of PC exposure to orange suppressive light, WM responses significantly decreased from 4.7 to 1.07 Hz at P13 and to 1.8 to 1.07 Hz at P21 (unpaired t test; R2 = 0.2432, p < 0.0001, and R2 = 0.1621, p < 0.0001), mirroring the frequency baselines and its reduction observed in Hx mice (Fig. 4D). Using the aforementioned spike templates from WM GABAergic interneurons and OPCs to profile the spike waveforms (Fig. 2C,D), we characterized WM responses during this period of reduced PC activity. Finally, similarly to Hx conditions, analysis of low (below 20 Hz) and high (20–60 Hz) firing frequency heat maps revealed that at P13, the contribution of OPCs to WM activity was higher than GABAergic interneurons (70.18 and 29.82%, respectively), while by P21 OPCs fully contributed to WM activity (Fig. 4E).
Chronic reduction of PC activity causes locomotor malfunction in normal mice and delays oligodendrocyte maturation and myelination in cerebellar WM
To assess the functional impacts of chronic PC activity reduction on locomotor performance in Pcp2-cre mice, we exposed Nx mice to the PC activity reduction protocol at P13 and P21 (Nx-muted) and compared them to Hx mice using two behavioral tests. Both the inclined beam climb test and rotarod test revealed that the behavioral phenotype of Nx-muted mice mimicked that of Hx mice. Specifically, on the inclined beam test, the Nx-muted and Hx mice exhibited significantly higher numbers of foot slips and longer completion times compared with Nx animals (one-way ANOVA, p < 0.0001; R2 = 0.6473 and R2 = 0.3009 on 1 cm beam, R2 = 0.5963 and R2 = 0.5192 on 2 cm beam; Movie 1, 2; Fig. 5A–D). Similarly, motor coordination assessment using the rotarod test showed a shorter latency to fall for Nx-mutated and Hx as compared with Nx. The Nx-muted group displayed an intermediate performance between Nx and Hx (one-way ANOVA, p < 0.0001; R2 = 0.4390; Fig. 5E). Importantly, no significant differences were observed in any of the measurements when we compared the Nx-muted and Hx mice, excluding the latency to fall time measured. These results demonstrate that reducing PC activity induces significant behavioral abnormalities and mimics the behavioral effects caused by Hx.
Suppression of PC activity mimics the effects of Hx on locomotor performance and a reduction in oligodendrocyte maturation and myelination. A, After the suppression of PC activity (green bar) and in Hx mice (pink bar), the average number of foot slips was significantly higher than in Nx mice (black bar; n = 10/group; 1-cm-wide beam). B, Similar results were obtained on a 2-cm-wide beam. C, D, Graphs represent the time to complete the task. E, On the rotarod test, locomotor performance was quantified as time latency to fall (seconds). In Nx mice with suppressed PC activity, time latency values are widely distributed over a range, with the lower average time comparable to normal Nx mice, whereas Hx-induced mice performed poorly with an average of only 100 s on the rotarod. F, Representative images of cerebellarWM from Nx, Hx, and Nx-muted mice, in which PC activity was suppressed. Tissue sections were immunostained at P25 with anti-NG2 for OPCs and anti-CC1 for mature OLs. DAPI was used for total cell counts and FluoroMyelin for myelin labeling. G, Quantification of the number of CC1+ and NG2+ cells in WM indicates that there was an approximately 50% reduction in CC1+ cell density in the Nx mice group in which PC activity was suppressed, as compared with Nx mice. This resembled the reduction in CC1+ cells observed in Hx mice. Conversely, a significant increase in NG2+ cell density was observed in Nx mice after PC activity reduction, similar to Hx mice. H, Changes in the ratio of CC1+ (gray) to NG2+ (pink) cell density indicate a reduction in the number of mature oligodendrocytes at P25 in the Nx mice with suppressed PC activity, as compared with Nx. I, The green mean fluorescent intensity plots show a 50% reduction in FluoroMyelin expression in WM after suppression of PC activity, similar to Hx mice. Scale bar, 100 µm. The effects of PC inhibition showed no difference in glial cell response and GABAergic cell numbers as indicated in Extended Data Figure 6-1. EM analysis, as shown in Extended Data Figure 5-1A,B, demonstrated a thicker myelination in Nx than in PC inhibited Nx-Muted mice, which further demonstrates that PC activity influences myelination.
Figure 5-1
Electron Microscopy Analysis of Myelin Structure in Nx and Nx Muted Mice. (A) Representative electron micrographs (50% zoomed in) of myelinated axons in the cerebellum under different conditions: normoxic (Nx, n = 3), Nx with Purkinje cells inhibited (Nx Muted, n = 2). Arrows indicate areas of disrupted or split myelin layers, characteristic of Hx myelination. Red arrows denote regions of dense, redundant myelin sheath degeneration with vacuoles and dense cytoplasm, also observed in Nx Muted. (B) Quantification of the g-ratio (inner diameter/outer diameter of myelinated axons), demonstrating thicker and denser myelination (lower g-ratio) in Nx compared to Nx Muted. These findings, along with the qualitative observations of myelin structure, support the conclusion that Purkinje cell activity significantly influences myelination in Nx conditions. (C) The percent un-myelination showed higher numbers of unmyelinated PCs Nx Muted mice as opposed to Nx mice. (D) Scatterplot of g-ratio versus axonal diameters in both Nx and Nx Muted mice groups suggesting relatively consistent myelination across axons with different diameters, with some variability. Each group included randomly picked (>60) axons with multiple fields of view for each animal. Correlation between axon diameter and g-ratio were presented as logarithmic regression curves (black) and correlation co-efficient (R2). Nx PC inhibited mice showed higher g-ratio (0.62) on relatively smaller axonal diameter (121.54 ± 49.8 s.d.) reducing the correlation regression from 0.442 to 0.343 (Nx axonal diameter was 165.44 ± 81.3 s.d.), indicating an overall thinner myelination in Nx Muted group. Download Figure 5-1, TIF file.
Following behavioral and locomotor experiments, we also investigated the effects of PC activity reduction on cellular properties of WM development. Mice from all three groups (Nx, Hx, and Nx-muted) were killed at P21, and cerebellar sections were immunostained for either OPCs (anti-NG2 antibody) or mature OLs (anti-CC1 antibody) and counterstained with DAPI (Fig. 5F). Cellular quantification of NG2+ and CC1+ cells (represented as percentages of total DAPI+ cells) revealed that in the Nx-muted group, the percentage of CC1+ cells decreased by approximately 50% as compared with Nx, but it was still higher than in the Hx group ∼ 50% (one-way ANOVA; R2 = 0.830, p < 0.0001; Fig. 5G). Conversely, both Nx-muted and Hx mice (one-way ANOVA; R2 = 0.6216, p < 0.0001) showed a significant increase in NG2+-expressing OPCs. The NG2+/CC1+ cell ratio (Fig. 5H) mirrored the reduction in OL maturation observed in Hx, indicating a similar cellular phenotype in Nx-muted mice. We also demonstrated that suppressing PC activity in Nx did not affect the expression of astrocytes or activated microglia (Extended Data Fig. 6-1A–C). Finally, staining of tissue sections with FluoroMyelin (Fig. 5F,I) demonstrated a reduction in myelin formation in Hx mice at P21 as compared with Nx (one-way ANOVA; R2 = 0.06777, p < 0.0001). This was confirmed by electron microscopy (EM) analysis, which showed decreased myelination in PC-suppressed Nx mice (Extended Data Fig. 5-2A–D). In summary, both cellular WM development and behavioral testing of Nx-muted mice indicated a similar phenotype to that of Hx-induced mice.
Chronic stimulation of cerebellar PCs in Hx mice promotes OL maturation and partially rescues locomotor abnormalities
To elucidate the pivotal role of PC activity in locomotor performance in Hx mice, as well as OL maturation, and developmental myelination in WM, we performed a gain-of-function experiment to reverse the effects of Hx on motor function. Following chronic optogenetic stimulation of PCs in Hx mice using a blue LED (λ = 473 nm) for 7 consecutive days, we assessed motor function at P21 using inclined beam and rotarod tests.
On the inclined beam test, Hx-stimulated mice had a significant reduction in the number of foot slips on both the 1- and 2-cm-wide incline beams, as compared with Hx mice (Fig. 6A,B). Notably, the number of foot slips on the 2-cm-wide beam was comparable with the Nx group. Furthermore, when compared with Hx mice, Hx-stimulated mice displayed a reduced time required to complete the task on the 2-cm-wide beam (one-way ANOVA; R2 = 0.2863, p < 0.0001; R2 = 0.6082, p < 0.0001; Fig. 6C,D), and when compared with Nx mice, the time Hx-stimulated mice required to complete any task was only significantly longer on the 1-cm-wide beam. Additionally, the latency to fall on the rotarod test significantly improved (one-way ANOVA; R2 = 0.4303, p < 0.0001) in Hx-stimulated mice as compared with the Hx group (Fig. 6E).
Selective chronic stimulation of PCs in Hx mice partially rescues locomotor malperformances and promotes OL maturation and myelination after Hx. A–D, Inclined beam test. Graphs represent the number of foot slips on a 1- and 2-cm-wide inclined beam and the time taken to complete the test. In both cases, a significant improvement was observed in the Hx-stimulated group, as compared with Hx mice. On a 2-cm-wide inclined beam, no significant difference was observed in the number of foot slips between Hx-stimulated and Nx mice. E, Rotarod test. Hx-stimulated mice displayed a significant improvement, as compared with Hx mice, although a complete rescue was not observed, when compared with Nx mice. F, Representative confocal images of cerebellar white matter brain slices from Nx, Hx, and Hx-stimulated mice. WM immunostaining with anti-NG2, anti-CC1, DAPI, and FluoroMyelin antibodies. G, Bar graphs represent the percentages of CC1+ and NG2+ cells in Nx, Hx, and Hx-stimulated mice. There is no significant difference in CC1+ mature OLs and NG2+ OPCs between Hx-stimulated and Nx mice. In contrast, the Hx mice displayed a reduction in CC1+ cells and an increase in NG2+ cells. H, The same trend was also observed in the ratio between CC1+ (gray) and NG2+ (pink) cells. I, The mean green fluorescent intensity plot from FluoroMyelin staining showed a significant increase in myelination in Hx-stimulated mice, as compared with Hx mice, with a very similar CTCF expression to Nx mice. Scale bar, 100 µm. PC stimulation in Hx mice showed a difference in glial cell response and GABAergic cell numbers as indicated in Extended Data Figure 6-1. EM analysis, as shown in Extended Data Figure 6-2A,B, demonstrated a thicker myelination in Hx-stimulated mice than in Hx mice, which further demonstrates that PC stimulation increases myelination.
Figure 6-1
Effects of Purkinje Cell Manipulation on Glial Cells and GABAergic Neurons. This figure illustrates the effects of optogenetic PC inhibition and stimulation on glial cell responses and GABAergic neuron numbers in Nx and Hx conditions, respectively. Representative immunofluorescence images stained with (A) Calbindin (green)/DAPI (blue) were used to quantify GABAergic neurons, while (B) IBA1 (magenta)/DAPI (blue) an (C) GFAP (red)/DAPI (blue) staining were used to assess microglia and astrocyte activity, respectively (20X magnification). Higher magnification (100X) images reveal detailed cellular morphology of PCs, microglia, and astrocytes. Statistical significance is indicated as * < 0.05, ** < 0.01 compared to Hx controls. Nx: Normoxia, Hx: Hypoxia, PC: Purkinje cell, GFAP: Glial fibrillary acidic protein, IBA1: Ionized calcium-binding adapter molecule 1. Scale bar, 50 μm. Download Figure 6-1, TIF file.
Figure 6-2
Electron Microscopy Analysis of Myelin Structure in Hx and Hx Stimulated Mice. (A) Representative electron micrographs (50% zoomed in) of myelinated axons in the cerebellum under different conditions: hypoxic (Hx, n = 3), and Hx with Purkinje cells stimulated (Hx Stimulated, n = 3). Arrows indicate areas of disrupted or split myelin layers, characteristic of Hx myelination. Red arrows denote regions of dense, redundant myelin sheath degeneration with vacuoles and dense cytoplasm, also observed in Hx. (B) Quantification of the g-ratio (inner diameter/outer diameter of myelinated axons), demonstrating thicker and denser myelination (lower g-ratio) in Hx Stimulated compared to Hx. These findings, along with the qualitative observations of myelin structure, support the conclusion that Purkinje cell activity significantly influences myelination in Hx conditions. (C) The percent un-myelination showed higher numbers of unmyelinated PCs Hx mice as opposed to PC stimulated Hx mice. (D) Scatter plot of axon diameter against g-ratio in Hx and Hx Stimulated mice. In both groups, more than 60 axons were randomly picked with multiple fields of view. Correlation between axon diameter and g-ratio were presented as logarithmic regression curves (black) and correlation co-efficient (R2). The correlation regression dropped from R2 = 0.556 to R2 = 0.438 in Hx Stimulated v Hx mice, as a result of improvement in myeline thickness (0.64 to 0.60) and increased average axonal diameter from 121.12 ± 65.3 s.d. to 149.02 ± 89.5 s.d. Download Figure 6-2, TIF file.
This clip shows Nx mice climbing the inclined beam (45°; 1 mm wide) in an efficient manner with minimum amounts of foot slips (1 time). There were no signs of impairment or miscoordination between the front and hind legs. [View online]
This clip shows the locomotion of Nx-muted mice on an inclined beam (45°; 1 mm wide). The animal showed hesitation in climbing the beam. It was significantly slower than the Nx mice, was dragging its hind leg, and had lots of foot slips during the climb, indicating impairment in locomotion. [View online]
Immunohistochemical analysis of Hx-stimulated mice at P21 revealed positive effects of stimulation on mice that had suffered Hx-induced WM injury (Fig. 6F). In Hx-stimulated mice, the percentage of CC1+ cells was higher (one-way ANOVA; R2 = 0.7052, p < 0.0001) than in the Hx group, and not different from the Nx mice (Fig. 6G). Additionally, the percentage of NG2+ cells in the Hx-stimulated mice was decreased (one-way ANOVA; R2 = 0.7551, p < 0.0001) as compared with the Hx group and was comparable to that of Nx mice. The ratio of mature to immature OLs in Hx-stimulated mice was reversed as compared with Hx mice and was practically identical to the Nx group (Fig. 6H). PC stimulation in Hx attenuates both astrocytosis and activated microglia (Extended Data Fig. 6-1A–C) suggesting that the restoral of cerebellar behavioral function upon PC stimulation might be partially mediated through PC-regulated astrocytosis and microglial activation. Finally, there was an increase in the mean fluorescence intensity of FluoroMyelin staining in Hx-stimulated mice as compared with the unstimulated Hx group (one-way ANOVA; R2 = 0.1008 p < 0.000001, Fig. 6I). The EM analysis further validated this finding as the PC stimulated Hx group showed a lower g-ratio count, indicating increased axonal myelination (Extended Data Fig. 6-2A–D). Interestingly, the FluoroMyelin intensity of the Hx-stimulated mice was very similar to that of Nx mice (Fig. 6I).
In summary, chronic PC stimulation after neonatal Hx not only restored OLs and myelin phenotypes in the developing cerebellar WM, but most importantly reversed locomotor deficits and improved function.
Discussion
We previously demonstrated that neonatal Hx induces functional impairment in cerebellar PCs in a clinically relevant mouse model of neonatal brain injury and linked abnormalities in PC physiology with specific aspects of locomotor dysfunction (Sathyanesan et al., 2018, 2021). In the present study, by selectively modulating PC activity in normal healthy mice, or by rescuing PC function after Hx, we uncover the regulatory influence of cerebellar PC inputs on oligodendrocyte maturation and myelination and their consequential impact on locomotor performance. In uninjured mice under normal physiological conditions, we observed both excitatory and inhibitory responses in WM cells in response to PC-evoked activity. The characterization of these WM responses revealed distinctive firing patterns in GABAergic interneurons, which displayed fast firing rate action potentials (APs; Galarreta and Hestrin, 2002; Tepper et al., 2010), and in NG2+ OPCs, which exhibited a slow firing rate and longer duration APs (Aguirre et al., 2004; Káradóttir et al., 2008). These differential responses suggest the involvement of GABA receptor-mediated hyperpolarization of GABAergic interneurons and OPCs, which are believed to play a crucial role in glial migration and differentiation during cerebellar WM development (Zonouzi et al., 2015).
Notably, our findings in PC-mediated activation of GABAergic interneurons and OPCs align with a specific developmental timeline, further emphasizing the impact of inhibitory interneuron migration on OPC differentiation into mature oligodendrocytes. This is underscored by our finding of increased OPC contribution in WM excitatory responses at P21 over P13 under normal conditions (Nx). Previous analysis with whole-cell patch-clamp recordings from purified cortical OPC cultures showed that accumulation of intracellular sodium caused by membrane depolarization inhibits OPC proliferation (Knutson et al., 1997). Our recent electrophysiological analysis of the mature WM showed that inhibitory interneurons migrate out of the WM into the cerebellar cortex by P20, and at that age, OPCs have differentiated into mature oligodendrocytes (Galas et al., 2017).
Furthermore, PC stimulation after Hx not only promoted oligodendrocyte maturation and myelination but also partially rescued Hx-induced locomotor malperformance, identifying PCs as critical regulators of oligodendrocyte development and locomotor performance. We demonstrated that Hx reduces spontaneous WM activity at P21, likely attributable to larger contributions of slow-firing OPCs over GABAergic interneurons. These results suggest that altered neuronal modulation of myelination contributes, at least in part, to the observed locomotor deficits induced by neonatal cerebellar injury during the early stages of cerebellar WM myelination. These findings align with previous studies that highlighted the crucial role played by proper development of the cerebellar cortex in motor control and in fine-tuning of motor coordination during early development and adolescence, as a disruption of this process results in longstanding abnormal motor function at later developmental stages (Manto and Jissendi, 2012). Importantly, we previously demonstrated that treatment with the GABA reuptake inhibitor tiagabine partially rescued locomotor dysfunction but did not affect adaptive cerebellar learning. This suggests that elevating extracellular GABA levels immediately following injury had positive developmental effects on the cerebellar neuronal circuitry that modulates proper WM development (van Welie et al., 2011; McKenzie et al., 2014; Sathyanesan et al., 2018).
WM repair and regeneration have been demonstrated in the early postnatal brain and after adolescence (Miyamoto et al., 2013). Mature, myelin-forming mature oligodendrocytes in WM are key players in locomotor performance (Ortiz et al., 2019), as OPCs rapidly proliferate and migrate to the affected areas, regenerating mature oligodendrocytes and restoring myelin integrity (Redwine and Armstrong, 1998; Yong, 2009; Lo, 2010). Several factors are critical for promoting full functional recovery, including the extent of injury, the potential for endogenous repair, and the time window of treatment or intervention (Lo, 2010). WM lesion studies in rats have highlighted a robust regenerative response of OPCs and preoligodendrocytes immediately after injury, which compensates for the lost immature oligodendrocytes. However, despite the presence of numerous OPCs at the injury site, the arrest of their differentiation and maturation impairs full regeneration and recovery (Back et al., 2002; S. Lin et al., 2004; Segovia et al., 2008). Our loss-of-function approach in Nx mice involved reducing PC activity and input to the WM, which mimicked the cellular and developmental effects observed after Hx. Under these conditions, we found a reduced rate of OPC maturation comparable to that observed in Hx mice, as indicated by the NG2+/CC1+ cell ratio. We also observed a reduction of approximately 50% in developmental myelination, accompanied by locomotor abnormalities like those observed in Hx mice. These results provide in vivo evidence that PC activity not only modulates cellular dynamics in WM and influences OPC differentiation to mature oligodendrocytes but also affects locomotor performance. Previous studies underscored the beneficial effects of expanding the pool of mature oligodendrocytes and enhancing myelination of the corpus callosum, which has led to improved motor neuron activity after demyelination (Bechler et al., 2018; Ortiz et al., 2019). In our gain-of-function experiments with Hx mice, we continuously applied blue light stimulation to ChR2-expressing PC for 7 d, which resulted in significant improvement in locomotor performance. While the performance of Hx mice with PC stimulation did not reach parity with their Nx counterparts in every test, our results demonstrate that controlled and persistent stimulation of PCs immediately after WM injury partially rescues abnormalities in locomotor coordination.
We provide the first in vivo evidence of crosstalk between PCs, OPCs, and GABAergic interneurons in the developing cerebellar WM. Our findings emphasize the distinct contributions of each cell type to myelin plasticity and functional recovery after neonatal brain injury. Notably, our study highlights the critical influence of PC activity on oligodendrocyte maturation, myelin formation, and locomotor coordination. Our analysis represents a first step to identify intrinsic and extrinsic factors that modify oligodendrocyte regeneration and maturation after cerebellar WM injury in neonates due to Hx injury in their NICU course. The potential therapeutic impacts of neuromodulation targeting PCs and other developing cerebellar cell types hold significant promise for promoting functional recovery. Techniques such as voltage- and current-controlled neuronal stimulation (ENS) or optical/magnetic/acoustic stimulation offer novel opportunities to develop new therapeutic interventions that support central nervous system recovery after injury or disease (Williams and Constandinou, 2013; Luan et al., 2014; Powell et al., 2019; Zhu et al., 2020).
Study Limitations
While our study provides valuable insights into the complex interplay between PCs, oligodendrocytes, and GABAergic interneurons in cerebellar WM development, there are limitations that merit some consideration. A primary limitation lies in our reliance on optogenetic modulation, which may evoke cellular responses beyond the physiological range. While this approach allows precise control over PC activity, which was critical for our study, the potential for off-target effects or nonphysiological responses should be acknowledged. To minimize that possibility, we utilized validated tracings from previous studies (Guo et al., 2010; D. Fan et al., 2012; H. Fan et al., 2017). Additionally, we utilized electrophysiological recordings for the classification of neuronal and oligodendrocyte types. The complex nature of cell subtypes suggests that a more nuanced characterization, for example utilizing molecular and genetic markers, may offer more comprehensive stratification. Future investigations incorporating multimodal approaches to both modulation and cell classification could enrich the interpretation of observed effects. Furthermore, in this study, we focused on a specific developmental time window (P13–P21); this focus allowed us to investigate a critical period in cerebellar maturation but does limit conjecture on the long-term consequences and persistence of the observed effects into adulthood. Expanding the study's temporal scope could provide a more relevant conceptualization of the enduring impact on cerebellar WM maturation. Finally, although the mouse model offers translational relevance, it also introduces species-specific considerations. Differences in cerebellar development between mice and humans should be acknowledged, and complementary studies in larger animal models or human tissues will be the critical next steps.
Footnotes
This work was supported by R37NS109478 (Javits Award; V.G.), K12HD001399 (P.K.), American SIDS Institute (P.K.), Raynor Cerebellum Project (P.K.), R21NS135088-01 (I.K.), 1K08NS131529 (T.D.), and R41AI167224-01A1 (I.K.). This study was also supported by the District of Columbia Intellectual and Developmental Disabilities Research Center (DC-IDDRC) Award P50HD105328 (V.G.) from the National Institute of Child Health and Human Development. The abstract figure of this paper was designed and made by Therese Mary Fischer.
The authors declare no competing financial interests.
- Correspondence should be addressed to Panagiotis Kratimenos at pankratimenos{at}gwu.edu or Vittorio Gallo at vittorio.gallo{at}seattlechildrens.org.
