Developmental Profiling of Structural and Functional Maturation in Mouse Corpus Callosum

ABSTRACT

As the largest white matter tract within the central nervous system (CNS) to connect two cerebral hemispheres, the corpus callosum axon bundle consists of a mixture of myelinated and unmyelinated axons and plays a crucial role in executing sensory, motor and cognitive functions within the CNS. In this study, we comprehensively characterized progressive alterations in myelination and oligodendrocyte lineage cell densities during the postnatal myelin development and then correlated these structural dynamics to the maturation of axonal impulse conduction within the mouse corpus callosum. In addition, we found that the extracellular spaces between callosal axons were significantly reduced during the first three postnatal weeks in mice, while micron‐scale diffusion of small molecule within this region remained largely unaffected and displayed isotropy. However, the glutamate transporter GLT‐1 was markedly upregulated within the first 3 postnatal weeks, and its expression was found not only in astrocytes but also in oligodendrocyte lineage cells. Finally, we showed that the ectopic callosal axonal vesicle machinery were not fully matured until the later state of myelin development. In summary, our study provided a dynamic profile of the structural and functional maturation of mouse corpus callosum during postnatal myelin development.

• Progressive myelin development in mouse corpus callosum is accompanied by gradual decrease in oligodendrocyte precursor cell density and increase in oligodendrocyte density.

• Pro‐Capase‐3 is transiently upregulated in oligodendrocyte during postnatal development.

• Impulse conduction along callosal axons is not optimized until later stages in myelin development.

• Extracellular space between callosal axons is substantially reduced during the first three postnatal weeks.

• Micron‐scale diffusion of small molecules within mouse corpus callosum is isotropic and remains unchanged during myelin development.

• The upregulation of glutamate transporter and the progressive maturation of axonal vesicle release machinary occur concurrently during callosal myelin development.

1. Introduction

The corpus callosum is the axon tract that connects the two cerebral hemispheres. Corpus callosal projection neurons are distributed throughout all cortical layers, with a higher abundance in layer II/III (Fame et al. 2011). As the largest white matter tract in the central nervous system (CNS), callosal axons transmit critical signals for sensory, motor and cognitive functions (Fenlon and Richards 2015). Myelination of these axons enhances axonal conduction and thereby can facilitates interhemispheric communication.

Myelin development primarily occurs postnatally. In mice, CNS myelination is not completed until postnatal day 60 and continues undergoing remodeling during adulthood (Young et al. 2013; Fields 2015; Hughes et al. 2018). Postnatal myelin development begins with robust proliferation of oligodendrocyte precursor cells (OPCs) and oligodendrogenesis via OPC differentiation during the first postnatal week and peaks during the second postnatal week (Rivers et al. 2008; Nishiyama et al. 2021). These differentiated oligodendrocytes then gradually initiate myelin ensheathment around axons. Early histology studies showed that in rodents, only 30%–40% of adult callosal axons are myelinated (Sturrock 1980; Mack et al. 1995). The mixture of myelinated and unmyelinated callosal axons is also supported by electrophysiology data. Compound action potentials (CAPs) recorded from the corpus callosum often display two separated peaks: a faster peak corresponding to myelinated axons, and a slower component representing unmyelinated axons (Preston et al. 1983; Silberstein et al. 1992). However, how CAPs in mouse corpus callosum evolve and mature during postnatal development is unclear and data on longitudinal CAP recordings across myelin development stages are limited.

Myelination is a highly plastic process influenced by many extrinsic factors. A number of extracellular signals, including neurotransmitters, ATP and adenosine, growth factors, cytokines, and hormones have been reported to modulate OPC proliferation and differentiation, and the subsequent myelin sheath formation (Hill and Nishiyama 2014; Bergles and Richardson 2015). From the early postnatal stage to adulthood, the corpus callosum, along with other brain regions, undergoes massive volumetric expansion. During the same time period, callosal projection refinement (Wang et al. 2007; De Leon Reyes et al. 2019, 2020) and myelin development are also occurring. These concurrent changes potentially alter the physical properties of the extracellular space (ECS) within the corpus callosum. However, the dynamic alteration in ECS structure and the subsequent impact on extracellular molecule diffusion within mouse corpus callosum have never been comprehensively studied.

Neuron–neuron synapses are exceedingly rare within the corpus callosum white matter. However, glutamate can still be released by callosal axons through discrete axonal vesicle release machinery (Kukley et al. 2007; Ziskin et al. 2007). Only a fraction of these vesicle release sites forms direct synaptic contacts with OPCs, whereas many of them do not show identifiable contact partners (Kukley et al. 2007). Glutamate signaling is considered an important player for myelin plasticity and myelin repair (see review by (Spitzer et al. 2016)). However, the developmental processes that govern axonal glutamate release, glutamate uptake by glutamate transporters and synaptic communication between axons and OPCs within the corpus callosum have not been studied.

In this study, we systematically characterized the dynamic progression of myelin formation and oligodendrocyte lineage cell densities in mouse corpus callosum across multiple stages of postnatal myelin development. We also performed electrophysiology recording to correlate the structural progression in myelination with the progressive improvement in axonal conduction function. In addition, we employed novel confocal shadow imaging techniques to visualize the ECS in mouse corpus callosum for the first time and optically analyzed molecule diffusion across different developmental time points. Finally, we quantified glutamate transporter expression and assessed the density of synaptic vesicles across different stages of myelin development. Taken together, our study utilized a multi‐modal approach to provide a comprehensive analysis of the structural and functional maturation of the mouse corpus callosum during postnatal myelin development.

2. Materials and Methods

2.1. Animals

For all listed experiments, female and male wild‐type (C57BL/6J, The Jackson Laboratories, Stock #000664) or Aldh1l1‐eGFP mice (Yang et al. 2011) (gifted by Dr. Min Zhou from The Ohio State University) were used around three timepoints: postnatal (P) days 10 (mice age ranges from P9 to P12), 20 (mice age ranges from P19 to P23), and 50 (mice age ranges from P48 to P54). Mice were randomly assigned to experimental groups. All experimental procedures were performed in line with IACUC guidelines at The Ohio State University.

2.2. Immunohistochemistry

Mice were deeply anesthetized with ketamine (150 mg/kg body weight) and xylazine (15 mg/kg body weight) and transcardially perfused with 4% paraformaldehyde (PFA). Brains were dissected and post‐fixed in 4% PFA for 1 h then stored in 1× phosphate buffered saline (PBS) before undergoing 10%, 20%, and 30% sucrose in 1× PBS graduations for cryoprotection. We embedded brain samples in Tissue‐Tek O.C.T. Compound and sliced the frozen sections at 20 μm thickness with a HM525 NX cryostat (ThermoFisher). Coronal brain sections containing the dorsal corpus callosum spanning the septal to rostral hippocampal levels (AP~ +0.8 mm to −1.7 mm relative to bregma in P50 mouse brains) were processed. After heating slides at 30°C for 30 min and three 10‐minute washes with 1× PBS on an orbital shaker, nonspecific binding was blocked with 2.5% BSA and 0.1% Triton in 1× PBS for 1 h at room temperature. Following PBS washes, the samples were incubated in primary antibodies diluted in 1× PBS with 1.25% BSA and 0.05% Triton overnight (for PDGFRα antibody, 24 h) at 4°C. The following morning, sections went through three 10‐minute PBS washes and were incubated with Alexa Fluor‐conjugated secondary antibodies (Life Technologies) at 1:400 dilution in 1× PBS for 1 h at room temperature. Fluorescence images were taken using an inverted epifluorescence microscope (Axio Observer Z1, Zeiss) or laser‐scanning confocal microscope (C2 plus, Nikon). For imaging analysis, 3–5 scans were acquired and analyzed for each animal.

For sections stained with FluoroMyelin Green (Invitrogen, Cat # F34651), after heating up the frozen sections and PBS washes, we incubate the sections with diluted FluoroMyelin Green dye (1:250 in 1× PBS) for 40 min at room temperature. Then the sections were washed in 1× PBS for two times and stained with DAPI before being cover slipped.

For sections stained with cleaved Caspase‐3 antibody (Cell Signaling Technology, Cat #9661), an antigen retrieval step was performed prior to the blocking step. Sections were incubated in a citric acid‐based antigen‐unmasking solution (1:100 dilution with 1× PBS, Vector Laboratories, Cat # H‐3300‐250) in boiling water at 95°C–100°C for 5 min. They were kept in the same solution for 25 min at room temperature to cool down. Afterward, sections were washed three times with 1× PBS and proceeded to the blocking step following our standard staining protocol.

For Aldh1l1‐eGFP brain sections stained with anti‐GFP nanobody, GFP‐Booster Alexa Fluor 488 (1:400 dilution in 1× PBS with 1.25% BSA and 0.05% Triton, Proteitech, Cat #gb2AF488) was added together with other Alexa Fluor‐conjugated secondary antibodies (1:400, Life Technologies). Sections were incubated in the GFP nanobody and secondary antibodies mixture for 1 h at room temperature. Normal PBS washing, DAPI staining and cover slipping steps were performed after the incubation.

2.3. Immunoblotting

Wild‐type mice from P10, P20 and P50 age groups were euthanized with CO₂. Fresh, whole corpus callosum tissue was rapidly dissected on ice under a Zeiss Stemi 200 stereo microscope by cutting an approximately 2.5–3.0 mm thick coronal brain tissue block containing dorsal corpus callosum spanning the septal to rostral hippocampal levels (approximately spanning AP +1.0 mm to −1.7 mm relative to bregma in P50 mouse brains) with a razor blade (Fisher, Cat# 50‐949‐216) then separating the dorsal and ventral borders of the corpus callosum from the surrounding gray matter with Dumont No. 2 forceps (FST, Cat# 11223‐20). Tissue was then frozen on dry ice. RIPA buffer (0.5 M Tris‐HCl [pH 7.4], 1.5 M NaCl, 2.5% deoxycholic acid, 10% NP‐40, 10 mM EDTA) containing protease inhibitor cocktail (Sigma, Cat # 04693132001) and the phosphatase inhibitor tablet (Sigma, Cat # 4906845001) was added at a volume determined by the relative size of the tissue for a final protein concentration between 1 and 5 mg/mL. Tissue was manually homogenized using 1.5 mL disposable pellet pestle tissue grinders by pressing and twisting the tissue in solution repeatedly 15 times until fully homogenized. After sitting on ice for 30 min, the tubes were then centrifuged for 10 min at 13,000 rpm in a refrigerated microcentrifuge. Then, we aspirated the supernatant and transferred it to a fresh tube on ice and the pellet discarded. The concentration of protein in the resulting lysate was determined by a Bradford assay using the Bradford reagent (Bio‐Rad Cat # 5000006). Samples contained a final protein of 10 μg and were fractionated by 10% Mini‐PROTEAN TGX Stain‐Free Protein Gels (BioRad, Cat# 4568033) using the Mini Protean Tetra cell (BioRad, Cat # 1658004). The separated proteins were then transferred from the gel to a 0.2 μm nitrocellulose membrane (BioRad, Cat # 1620112) via a three‐hour transfer process at 300 mA using the Tetra and Blotting Module (BioRad, Cat # 1660827EDU). To confirm the equal loading, the membrane was stained with Ponceau S (Sigma, Cat # P7170). The membrane was washed with 1× Tris‐buffered saline with 0.1% Tween20 (TBS‐T) for 10 min before incubating in 5% nonfat milk in 1× TBS‐T blocking solution for 2 h at room temperature on a rocker. Following three 10‐minute washes with 1× TBS‐T, the membrane was treated with primary antibodies for 2 h at room temperature. After three TBS‐T washes, the membrane underwent a one‐hour incubation in Horseradish Peroxidase (HRP)‐conjugated secondary antibodies (1:10000 with TBS‐T, Goat anti‐Guinea Pig IgG (H+L) Highly Cross‐Adsorbed HRP conjugated Secondary Antibody, Invitrogen, Cat # A18775; Donkey anti‐Rabbit IgG (H+L) Cross‐Adsorbed HRP conjugated Secondary Antibody, Invitrogen, Cat # 31458) at room temperature. After washes with 1× TBS‐T and then distilled water, the membrane was treated with SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher Cat # 34580) for 5 min in dark at room temperature and exposed to CL‐XPOSURE film in a dark room and developed to visualize the protein bands. Films were scanned on a Xerox Altalink C8145 printer‐scanner at 600 DPI and quantified on ImageJ (NIH) image processing software. The mean gray value of the band was measured, background subtracted and normalized to the loading controls.

2.4. Antibodies

The following primary antibodies or nanobodies were used in this study:

Rabbit anti‐EAAT1 (GLAST) (1:5000 for immunoblotting, Abcam, AB416, RRID: AB_304334).

Guinea pig anti‐EAAT2 (GLT‐1) (1:5000 for immunoblotting, 1: 1000 for immunohistochemistry, Millipore, AB1783, RRID:AB_90949).

Goat anti‐PDGFRα (1:400 for immunohistochemistry, R&D Systems, AF1062, RRID: AB_2236897).

Mouse anti‐CC1 (1:400 for immunohistochemistry, Calbiochem, OP80‐100UG, RRID:AB_2057371).

Goat anti‐Caspase‐3 (1:200 for immunohistochemistry, R&D Systems, AF‐605‐SP, RRID: AB_354518).

Rabbit anti‐cleaved Caspase‐3 (Asp175) (1:400 for immunohistochemistry, Cell Signaling Technology, 9661, RRID: AB_2341188).

Rabbit Anti‐Olig2, clone 1K2 ZooMAb (1:600 for immunohistochemistry, Sigma‐Aldrich, ZRB1436, RRID: AB_3720355).

Rabbit anti‐VAMP2 (1:400 for immunohistochemistry, Cell Signaling Technology, 13508, RRID: AB_2798240).

Chicken anti‐mCherry (1:500 for immunohistochemistry, Abcam, AB205402, RRID: AB_2722769).

ChromoTek GFP‐Booster Alexa Fluor 488 (1:400 for immunohistochemistry, Proteitech, gb2AF488, RRID: AB_2827573).

The following secondary antibodies were used for immunoblotting:

Donkey anti‐Rabbit IgG (H+L) Cross‐Adsorbed Secondary Antibody, HRP.

(1:10000 for immunoblot, Invitrogen, 31458, RRID: AB_228213).

Goat anti‐Guinea Pig IgG (H+L) Highly Cross‐Adsorbed Secondary Antibody, HRP.

(1:10000 for immunoblot, Invitrogen, A18775, RRID: AB_2535552).

For all immunohistochemistry staining, we used Alexa Fluor‐conjugated (488, 555 or 647) secondary antibodies (1:400, Life Technologies).

2.5. ImageJ Imaging Processing

Fluorescence images of FluoroMyelin Green staining (Figure 1) were acquired with an inverted epifluorescence microscope (Axio Observer Z1, Zeiss). We acquired 3–5 images for each animal. Coronal brain sections containing the dorsal corpus callosum spanning the septal to rostral hippocampal levels were stained and processed. Images were taken with a 10× objective (NA 0.3, Zeiss) for an image area of 1249.41 μm × 1002 μm. The central portion of the corpus callosum, located directly beneath the cortical midline, was placed in the center of image. A region of interest (ROI) that outlines the corpus callosum region within the image was drawn for each image. We also drew another background ROI in the gray matter area in the cortex. The average fluorescence intensities of the two ROIs were measured, and we calculated the ∆fluorescence intensity by subtracting the average fluorescence intensity of the background ROI from the corpus callosum ROI average fluorescence intensity. ∆fluorescence intensity measurements from multiple scans for each animal were averaged to obtain the data points in Figure 1B.

FIGURE 1.

Progressive myelin formation in mouse corpus callosum from postnatal day 10 to 50. (A) Images showing FluoroMyelin Green and DAPI staining on mouse corpus callosum region at postnatal day 10, 20, and 50. Schematic inset shows the region where images were taken (blue dashed box). White dashed lines indicate how the corpus callosum was outlined for FluoroMyelin fluorescence intensity measurement. (B) Bar graphs illustrating the increase of FluoroMyelin signal within the corpus callosum, compared to background signal in the cortex, across three age groups. One Way ANOVA with post hoc Tukey's HSD test for multiple group comparisons. p < 0.001 for overall age group effect. ***p < 0.001. N = 4 mice for each age group. (C) Label‐free imaging of live tissue myelinated axons at P10, P20, and P50 using the spectral confocal reflectance microscopy (SCoRe). Top: Composite SCoRe images that combine reflective signals from 488, 561, and 641 nm channels. Bottom: Mask images demonstrating pixels with SCoRe signal intensities higher than the threshold (five times the background signal). (D) Bar graphs illustrating the percentage of myelinated area calculated from SCoRe images. Mixed effects models to test the group effect on outcomes with Tukey adjustment for multi‐group comparison. p < 0.01 for overall age group effect. *p < 0.05, **p < 0.01. N = 9 slices from P10, N = 10 slices from P20 and N = 11 slices from P50 group.

To quantify the densities of different cell types in the corpus callosum, immunostained brain sections were scanned with a laser‐scanning confocal microscope (C2 plus, Nikon) with a 20× objective (NA 0.95, water‐immersion, Nikon). 3–5 Z‐stacks with the z‐step size of 1.5 μm were acquired for each animal. OPCs, oligodendrocytes, Caspase‐3+ and cleaved Caspase‐3+ oligodendrocytes were counted using the ImageJ Cell Counter plugin. A ROI that outlines the corpus callosum region was drawn for each z‐stack. We counted the total cell numbers for the whole z‐stack and divided by the area of the ROI to obtain the density measurement.

We quantified VAMP2+ puncta density in the corpus callosum (CC) using the maximum intensity projection of the confocal z‐stacks. The threshold of each image was set by the experimenter in ImageJ. The VAPM2+ puncta were segmented with the ImageJ watershed tool after background thresholding. VAMP2+ puncta were then automatically identified with the “analyze particles” tool (range: 0.10‐infinity pixels) and divided by the area of the outlined corpus callosum region. For quantifying VAMP2+ puncta along AAV‐tdTomato transduced axons, we used the semi‐automatic tracing (SNT) plugin in ImageJ to trace visible tdTomato+ axons and manually counted visible VAMP2+ puncta that overlap with traces in the x, y and z axes using the Cell Counter plugin.

2.6. Corpus Callosum Axon Tracing

To trace callosal axons at P50, we anesthetized 5‐week‐old mice with a mixture of ketamine (100 mg/kg body weight) and xylazine (10 mg/kg body weight). Mice were placed on stereotaxic frame, and we performed a craniotomy to expose the right sensory‐motor cortex. Adeno‐Associated Viral (AAV) particles expressing the red fluorescent tdTomato protein (AAV‐CAG‐tdTomato (codon diversified), Addgene Cat # 59462‐AAV1, gift from Edward Boyden, RRID:Addgene_59,462, 1:3 dilution with ddH₂O from the original titer of ≥ 5 × 10¹² vg/mL) was injected into the right sensory‐motor cortex at 300–400 μm depth (AP coordinates from bregma in mm: AP +0.2/1.1, −1.0/1.1, 400 nL/injection site at 50 nL/min speed using Nanoliter 2010, World Precision Instruments). Animals were perfused at P50. Frontal frozen brain sections were prepared and stained with antibodies against mCherry and Vamp2.

To trace callosal axons at P10 and P20, neonatal wild‐type pups (P1‐P2) were cryoanesthetized with ice and injected with a total of 1.2 μL of AAV‐CAG‐tdTomato into the sensory‐motor cortex using the Nanoliter 2010 at the speed of 500 nL/min (two spots in the right hemisphere and 1 spot in the left hemisphere, 400 nL per spot, 500 nL/min speed). Pups were placed on a heating pad to warm up before being returned to the breeding cage. Animals were perfused at P10 or P20 and frozen sections were prepared for immunostaining.

2.7. Acute Brain Slicing

Wild‐type mice from P10, P20, and P50 age groups were anesthetized with Isoflurane. We quickly dissected out the brain and immersed it into the ice‐cold dissecting solution containing NaCl (87 mM), KCl (2.5 mM), NaH₂PO4 (1.25 mM), NaHCO₃ (25 mM), MgCl₂ (7 mM), CaCl₂ (0.5 mM), Sucrose (75 mM), Glucose (10 mM) (gassed with 95% O₂/5% CO₂). We then sliced brain tissue into 300 µm thick frontal slices using a vibratome (VT1200, Leica) in ice‐cold dissecting solution. After finishing slicing, we transferred the slices into dissecting solution at 35°C for 25 min of recovery. The brain sliced were then transferred into Artificial Cerebrospinal Fluid (ACSF, NaCl 124 mM, KCl 3 mM, NaH₂PO4 (1.25 mM), NaHCO₃ (26 mM), MgCl₂ (2 mM), CaCl₂ (2 mM), Glucose (10 mM), gassed with 95% O₂/5% CO₂) at room temperature and left there for at least 30 min before recording or imaging experiments.

2.8. Shadow Imaging

An acute frontal brain slice was placed in the imaging chamber with continuous perfusion of gassed ACSF on a motorized XY stage (Z Deck, Prior) for a fixed‐stage microscope (FN1, Nikon). A glass patch pipette loaded with 500 µM Alexa Flour 568 Dextran dye (10,000 MW, ThermoFisher, Cat # D22912) diluted in ACSF was attached to a pipette holder and positioned ~50 μm below the surface of the slice in the corpus callosum area using a micromanipulator (MPC‐200, Sutter Instrument). The glass pipettes were pulled using a vertical pipette puller (Model PC‐100, Narishige) with resistances around 4–6 MΩ. The dye was then puffed out with a constant small positive pressure using a 1 mL syringe. Red fluorescent images (excitation wavelength 561 nm, emission filter: 575–620 nm) were acquired using a laser‐scanning confocal microscope (AX system, Nikon) with a 60× water‐immersion objective (NA 1.0, Nikon). In the Nikon imaging software NIS‐Element, the imaging z‐stack were deconvolved using 3D deconvolution package (Landweber method, noise level and iterations adjusted depending on quality of original image). We then inverted the Look‐Up Table (LUT) to show ECS as dark regions. We performed all downstream imaging analysis in ImageJ (NIH). Three parallel lines with width of 1 pixel were drawn perpendicular to the direction of the axons, and their intensity profiles were taken. These profiles were then imported into Igor Pro (WaveMetrics) and we used the TN020‐B Peak Areas Macro (included in Igor Tech Notes directory) for further analysis: We subtracted the background fluorescence from the line profiles by fitting the baseline with a poly5 function and subtracting the baseline from the line profile using the TN020‐B macro. ECS between axons were identified in the macro as negative valleys and we measured the size of identified space widths as the Full Width at Half Maximum (FWHM). All identified negative valleys were visually inspected by the experimenter.

2.9. Dye Diffusion Imaging

A glass pipette (resistance around 4–6 MΩ) loaded with 50 μM Alexa Flour 594 dye (ThermoFisher, Cat # A10438) diluted in ACSF was positioned ~10 μm below the surface of the slice within the corpus callosum across all experiments. The dye was ejected with an electrical pulse (60 V, 6 ms bi‐phasic stimuli) generated by an Isolated Pulse Stimulator (Model 2100, A‐M Systems). The fluorescence image (excitation wavelength 561 nm, emission filter: 575–620 nm) capturing the dye diffusion was acquired in line‐scan mode using the Nikon AX laser‐scanning confocal microscope with a 60X water‐immersion objective (NA 1.0, Nikon). The line was placed across the pipette tip at either parallel or perpendicular direction of the callosal axons, and the frequency of the line scan was set at 1000–1500 Hz. The line‐scan was initiated 400 ms before the dye ejection and continued until 1024 lines were scanned, resulting in a total scan time of 800–1000 ms per image. For all collected data points across three age groups, we ensured that the brightest fluorescence intensity at the release site did not saturate our confocal detector.

All analyses were performed on 12‐bit raw grayscale images. From the line‐scan images, vertical line profiles (width of 0.5 μm) capturing time‐lapse fluorescence dynamics were taken in Igor Pro (WaveMetrics) at different distances from the pipette tip. The line profiles were background‐subtracted and smoothed using a sliding average of five points. We then normalized the peak amplitude of line profiles at each distance to the maximum fluorescence increase at the dye ejection site. The peak fluorescence amplitude of each normalized line profile was measured manually via placing the crosshair cursor for each line profile. Normalized fluorescence as a function of distance from the dye ejection site was fitted with an exponential decay function. The fitting yields a length constant λ, which we used as the indication of the optical spread of the fluorescence dye. We optimized our imaging configuration parameters to ensure a broad dynamic range between peak fluorescence intensity at the release site and background noise. The high signal to noise ratio allowed multiple fluorescence measurements across distances for reliable exponential fitting to obtain the estimates of diffusion length constant λ.

2.10. SCoRe Imaging

The corpus callosum region from the acute brain slice was scanned in Nikon AX laser‐scanning confocal microscope with the 60X water‐immersion objective (NA 1.0, Nikon) and an additional digital zoom of 3 or 4. The reflect signals, not fluorescence signals, were collected by photomultiplier detectors with the excitation of 488, 561, and 633 nm lasers. The reflective images from the three channels were then composited together in ImageJ to generate the SCoRe images for showing myelinated axons.

To quantify the myelinated area, we first measured the background intensity from a background ROI within the same image that lacked SCoRe signal. We then applied an intensity threshold set as five times the background signal. A 30 × 30 μm square was placed over the axon bundle region for each image. The percentage of myelinated area within this square was calculated by summing all pixels with intensity higher than the threshold and then dividing this sum by the total pixel number of the 30 × 30 μm square. For each acute brain slice, three SCoRe images were acquired at different depths ranging from 5 to 30 μm beneath the slice surface. The calculated percentages of myelinated area from these three images were then averaged for each brain slice.

2.11. Compound Action Potential (CAP) Recording

Recording and stimulating glass pipettes were pulled using the vertical puller (Model PC‐100, Narishige) with resistances of 1–2 MΩ. Both pipettes were loaded with ACSF. The recording pipette was first positioned in the corpus callosum just below the surface of the slice. We then placed the stimulating pipette with similar depth beneath the surface in the corpus callosum at three distances from the recording pipette: 500 µm, 1000 μm and 1300 or 1500 μm. CAPs were recorded at each distance with an electrical stimulation pulse (0.8 ms, single‐phase, 99 V, Model 2100, A‐M Systems). CAPs were recorded at above mentioned 3 distances using an NPI ELC‐03XS Universal Amplifier (External mode) with the gain set at 100×. Signals were sampled at 100 kHz using the WinWCP software (University of Strathclyde Glasgow). Following baseline recordings, we bath applied 1 μM tetrodotoxin (TTX) and repeated the recordings at the same three distances. All recording files were exported and analyzed in Igor Pro 9 (WaveMetrics). To isolate the CAP signal, we subtracted the post‐TTX recordings from the corresponding raw recordings. At 1000 μm, CAPs began to display more than one peak, reflecting axons bundles with different conduction velocities. From the isolated CAP signal, we defined the first downward peak that crossed the 0 mV baseline as the first peak, and any noticeable second downward peak following the first peak as the second peak. The latencies of first peak at 500 μm and 1000 μm were measured as the time difference between the onset of stimulation and the first peak of the CAP. To calculate the conduction velocity of the first and second peaks, we used recordings obtained at 1000 and 1300 or 1500 μm. The velocity was calculated by dividing the difference between two distances by the differences in CAP peak timings.

2.12. Statistical Analysis

All bar graphs show individual data points, means and standard errors of the mean. In the analysis with repeated measurements from one animal, we utilized mixed effects models to test the group effect on outcomes. Fixed effects included age groups, as well as direction, and their interaction for diffusion analysis. For other repeated measurements, only age group was included as fixed effect. Random effect included a subject‐specific random intercept to account for within‐subject correlation (repeated measurements). The Kenward‐Roger adjustment to the degrees of freedom was used to control Type I error rates. For data with only one measurement per animal, one‐way ANOVA models were used. Two‐sided significance of α = 0.05 was used with Tukey's adjustment for multiple group comparison. For the analysis of the percentage of CAPs recording at 1000 μm showing more than one peak across three age groups and the percentage of traced axons displaying co‐localization with VAMP2+ puncta across three age groups, we used Fisher's Exact Test or Chi‐square test depending on the sample size with Bonferroni adjustment for multi‐group comparison. With Bonferroni adjustment, we declare significance at 0.05/3 = 0.017. All analyses were performed in SAS version 9.4 (SAS institute, Cary, NC), except that the one‐way ANOVA with post hoc Tukey's HSD test for multi‐group comparison for Figures 1B and 7B was performed using R (version 4.4.2).

FIGURE 7.

The density of VAMP2+ vesicles markedly increased within the corpus callosum during postnatal development. (A) Immunostaining of synaptic vesicle SNARE protein VAMP2 in the corpus callosum. (B) Bar graphs show an overall increase in VAMP2+ puncta density within corpus callosum from P10 to P50. One‐way ANOVA with Tukey's HSD test for multi‐group comparison. p < 0.01 for overall age group effect. **p < 0.01. N = 7 for P10 group, N = 6 for P20 group, N = 8 for P50 group. (C) Representative single‐plane confocal scan showing immunolabeling of VAMP2 and callosal axons transduced by AAV‐tdTomato. Orthogonal projections show the co‐localization of VAMP2+ puncta on a tdTomato+ axon. (D) Histogram showing comparable lengths distribution from traced axons across the three age groups. (E) Stacked bar graphs illustrate the percentage of traced axons either lacking (open box) or co‐localized with VAMP2+ puncta (solid box). Chi‐square test for 3 samples with Bonferroni adjustment for multi‐group comparison. p < 0.01 for overall age group effect. With Bonferroni adjustment, we declare significance at 0.05/3 = 0.017. **p < 0.01. N = 198 for P10 group, N = 271 for P20 group, N = 411 for P50 group. (F) Cumulative probability plots showing puncta number per 100 μm across the three age groups from all traced axons that showed co‐localization with VAMP2+ puncta.

3. Results

3.1. Progressive Myelin Development in Mouse Corpus Callosum Is Accompanied by Gradual Decrease in OPC Density and Increase in Mature Oligodendrocyte Density

We started out with characterizing the myelin development progression in mouse corpus callosum via performing the FluoroMyelin Green stain on fixed mouse frontal brain sections at three postnatal time points: postnatal (P) day 10, 20 and 50 (Figure 1A). These three time points were chosen to represent the initial, middle and late phases of myelin development. Correspondingly, we observed a progressive increase in fluorescent myelin signal intensity: low at P10, elevated at P20, and markedly higher at P50 (Figure 1B). We then utilized a label‐free myelin imaging technique, Spectral Confocal Reflectance Microscopy (SCoRe) (Schain et al. 2014), to visualize myelinated axons in the corpus callosum from live tissues using acute brain slices. Consistent with Fluoromyelin staining results using fixed sections, in live tissue, we observed very few myelinated axons at P10, followed by an increased number at P20, and a further increase by P50 (Figure 1C). The quantification shows from P10 to P50, myelinated area within callosal axon bundles increased from 4.28% ± 2.8% at P10 to 49.80% ± 7.1% at P50 (Figure 1D).

To correlate the myelin development with oligodendrocyte lineage cell dynamics, we quantified the density of PDGFRα+ OPCs and CC1+ mature oligodendrocytes via immunohistochemistry (Figure 2A,C). OPC density gradually declined from 11.86 ± 1.3 to 4.97 ± 1.2 cells per 10,000 sq. μm between P10 and P50, representing a > 50% reduction (Figure 2B). In contrast, mature oligodendrocyte density increased from 8.18 ± 0.6 to 15.67 ± 0.7 cells per 10,000 sq. μm over the same period, representing a > 80% increase (Figure 2D). Previous reports have showed that a high fraction of developmental differentiating oligodendrocytes undergo programmed cell death in various CNS regions, including optic nerve, cerebral cortex and spinal cord (Barres et al. 1992; Trapp et al. 1997; Calver et al. 1998). To investigate the apoptosis machinery in oligodendrocytes within corpus callosum during the postnatal development, we first labeled brain sections from 3 age groups with antibodies against CC1 and Caspase‐3, which detects both cleaved Caspase‐3 and intact full‐length pro‐Caspase‐3 (Figure 2C). In P10 group, 27.88% ± 2.7% of CC1+ oligodendrocytes within corpus callosum were positive for Caspase‐3. This percentage increased to 40.34% ± 4.3% in P20 group. In P50 group, however, only 14.51% ± 0.5% of CC1+ oligodendrocytes were Caspase‐3+ (Figure 2E p < 0.001 for overall age group effect; p < 0.05 for P10 vs. P20; p < 0.001 for P20 vs. P50 and p < 0.05 for P10 vs. P50). Next, we labeled the brain sections from same groups of animals with antibodies against CC1 and cleaved Caspase‐3. Surprisingly, we did not observe any CC1+ oligodendrocytes showing immunoreactivity to cleaved Caspase‐3 (Figure 2F, N = 4 mice for each age group, 3 confocal z‐stack scans for each animal). Our results suggest that oligodendrocytes express high levels of pro‐Caspase‐3 during first 3 postnatal weeks without undergoing active apoptosis. This upregulation of pro‐Caspase‐3 is weakened in later developmental stage.

FIGURE 2.

Progressive myelin development is accompanied by gradual decline in oligodendrocyte precursor cell (OPC) density and increase in mature oligodendrocyte density. Pro‐caspase‐3 signal is elevated in oligodendrocytes during developmental stages but decreases upon reaching adulthood. (A) Single focal plane confocal images showing the immunostaining of PDGFRα and DAPI for identification of OPCs (white arrow heads). (B) Bar graphs illustrating the quantification of OPC density within the mouse corpus callosum at P10, P20, and P50. One‐way ANOVA with Tukey's HSD test for multi‐group comparison. p < 0.01 for overall age group effect. *p < 0.05, **p < 0.01. N = 4 mice for all three age groups. (C) Single focal plane confocal images showing the immunostaining of CC1, Caspase‐3 and DAPI for identification of oligodendrocytes (CC1+ and DAPI+) and Caspase‐3+ oligodendrocytes (CC1+, DAPI+ and Caspase‐3+, yellow arrow heads). (D) Bar graphs illustrating the quantification of oligodendrocyte density within the mouse corpus callosum at P10, P20 and P50. Mixed effects models to test the group effect on outcomes with Tukey adjustment for multi‐group comparison. p < 0.001 for overall age group effect. **p < 0.01, ***p < 0.001. Two sets of independent CC1 immunostaining and counting performed from N = 4 mice for all three age groups. (E) Bar graphs showing the percentage of Caspase‐3+ oligodendrocytes. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.001 for overall age group effect. *p < 0.05, ***p < 0.001. N = 4 mice for all three age groups. (F) Single focal plane confocal images showing the immunostaining of CC1, cleaved Caspase‐3 and DAPI for identification of apoptotic oligodendrocytes. Note that we did not observe any oligodendrocyte being cleaved Caspase‐3+ within corpus callosum (N = 4 mice for all three age groups). Light blue arrow heads point to other apoptotic cells that are cleaved Caspase‐3+ but CC1‐. White dashed lines outline the corpus callosum regions.

3.2. Axonal Conduction Velocity Shows Continued Refinement During Postnatal Myelin Development

To correlate the myelin formation dynamics at structural level with axonal conduction function, we recorded compound action potentials (CAPs) from callosal axons at the above‐mentioned three developmental time points. For each recording, the stimulating electrode was placed at three distinct distances from the recording electrode (Figure 3A): a close distance (500 μm), an intermediate distance (1000 μm) and a far distance (either 1300 or 1500 μm). To isolate the biological CAP signal from the stimulus artifact interference, we bath‐applied tetrodotoxin (TTX, 1 μM) after recording the initial stimulus‐evoked responses, and then repeated the recording protocol at the three distances. The CAP signal was obtained by subtracting the post‐TTX recordings from the corresponding pre‐TTX ones (Figure 3B). Across all age groups, the majority of CAPs recorded at 500 μm did not show separated peaks (Figure 3C). However, at 1000 μm, while 89% of CAPs recorded from the P10 group still only exhibited one peak, we observed that more than half of CAPs recorded from the P20 group exhibited more than one peak (Figure 3C,D). In the P50 group, 75% of CAPs recorded at 1000 μm exhibited a second peak (Figure 3C,D, p < 0.05 for overall age group effect, p < 0.017 for P10 vs. P50). We postulate that the increasing fraction of CAP recordings displaying two CAP peaks at later development time points is consistent with the progressive myelination of a larger population of axons. The faster CAP peak likely arises from myelinated axons, which are known to conduct more rapidly than unmyelinated axons, thereby contributing to the emergence of an earlier and faster CAP component.

FIGURE 3.

Compound action potential (CAP) recording illustrates the continued optimization in axonal conduction during postnatal myelin development in mouse corpus callosum. (A) Schematic showing stimulating and recording electrode placement for CAP recording. Recording electrode placement remained unchanged during recording, while the stimulating electrode was placed at 3 distances from the recording electrode, representing near (500 μm, #1), intermediate (1000 μm, #2) and far distances (1300 or 1500 μm, #3). (B) CAP signal (black solid line) was obtained by subtracting the stimulus‐evoked response after bath application of 1 μM tetrodotoxin (TTX, gray dashed line) from the raw recording before TTX application (gray solid line). Blue asterisk shows the onset of electrical stimulation. Scale bars for both recordings: 2 ms, 0.5 mV (C) Example CAPs recorded in mouse corpus callosum at 500 μm (left) and 1000 μm (right) from P10, P20 and P50 mice. Note that we only observed one CAP peak at 500 μm but frequently observed separated CAP peaks at 1000 μm from P20 and P50 mice. Arrows indicate the temporal location of the first peak for each recording and the color of arrow denotes the corresponding age group. The stimulation artifact was blanked out. Blue asterisk shows the onset of electrical stimulation. Scale bars: 1 ms, 0.3 mV (500 μm); 1 ms, 0.2 mV (1000 μm). (D) Stacked bar graphs illustrate the percentage of CAP recordings at 1000 μm either showing only one peak (open box) or exhibiting more than one peak (solid box). Fisher's Exact Test for 3 groups with Bonferroni adjustment for multi‐group comparison. p < 0.05 for overall age group effect. With Bonferroni adjustment, we declare significance at 0.05/3 = 0.017. *p < 0.017. N = 9 recordings for P10 group, N = 9 recordings for P20 group, N = 8 recordings for P50 group. (E) Bar graphs illustrating the latency measurements for the first peak at 500 μm. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.05 for overall age group effect. *p < 0.05. N = 9 recordings for P10 group, N = 9 recordings for P20 group, N = 8 recordings for P50 group. (F) Bar graphs illustrating the latency measurements for the first peak at 1000 μm. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.001 for overall age group effect. *p < 0.05. ***p < 0.001. N = 9 recordings for P10 group, N = 9 recordings for P20 group, N = 8 recordings for P50 group. (G) Bar graphs demonstrating the calculated conduction velocities for all identified first and second peaks. Mixed effects models to test the group effect on outcomes with Tukey adjustment. For the 1st peak, p < 0.01 for overall age group effect. *p < 0.05. For the 2nd peak, p > 0.05 for overall age group effect. N = 9 for P10 group, N = 9 for P20 group, N = 8 for P50 group.

To test whether this is the case, we next compared the latency between the onset of electrical stimulation and the first peak of recorded CAPs at 500 and 1000 μm across three age groups. At both distances, the latency for the first peak gradually decreased from P10 to P50 (Figure 3E,F), suggesting overall increased conduction velocity for the first CAP peak from P10 to P50. Finally, we calculated the conduction velocity of identified first and second peaks using recordings from the intermediate and far distances. The conduction velocity was calculated by dividing the difference between distances by the difference in peak timing. The conduction velocities of the first peak for the P10 group (0.30 ± 0.03 m/s) were comparable to the P20 group (0.31 ± 0.02 m/s, p > 0.05 for P10 vs. P20, Figure 3G). However, the first peak velocity for the P50 group (0.43 ± 0.04 m/s) was faster than both P10 and P20 groups (p < 0.05 for P10 vs. P50 and p < 0.05 for P20 vs. P50, Figure 3G). Our latency and conduction velocity measurements suggest that first CAP peaks recorded in P50 group are likely contributed by conduction in myelinated axons. In contrast, the conduction velocities of the second and slower peak, often associated with unmyelinated axons, were comparable across age groups (p > 0.05).

3.3. Extracellular Spaces (ECS) Between Axons Within Mouse Corpus Callosum Were Substantially Reduced During Postnatal Myelin Development

During the transition from early postnatal developmental stage to adulthood, brain volume expansion, callosal projection refinement (Wang et al. 2007; De Leon Reyes et al. 2019, 2020), and myelination all occur within the same time period. Given these physical changes, we became interested in investigating how the ECS in mouse corpus callosum is altered between P10 and P50. We adopted a recently developed shadow imaging technique (Tonnesen et al. 2018; Dembitskaya et al. 2023) to directly visualize the extracellular fluid in acute frontal brain slices using the laser‐scanning confocal microscope system (details see Methods, Figure 4A). From the inverted shadow images, we clearly observed parallel axon bundles across all age groups (Figure 4A). We drew straight lines perpendicular to the axonal direction and calculated the extracellular space width (ECS width) between axons from the obtained line profiles (Figure 4B). The histograms of all measured ECS widths showed that P10 group ECS widths exhibited a wider spread and overall larger width sizes than P20 and P50 groups (Figure 4C). Meanwhile, the distribution of ECS widths was comparable between P20 and P50 groups (Figure 4C). The average ECS width in the P10 group was significantly larger than the P20 and P50 groups (Figure 4D, p < 0.05 for overall age group effect. p < 0.05 for P10 vs. P20 and p < 0.05 for P10 vs. P50). These findings suggest that there is a significant reduction in mouse corpus callosum ECS within the first three postnatal weeks. However, the continuing myelination from P20 to P50 has minimal effect on ECS between axons.

FIGURE 4.

Confocal shadow imaging reveals a sharp decrease in extracellular space between callosal axons from postnatal day 10–20. (A) Single plane confocal inverted shadow image of extracellular spaces in the corpus callosum at P10, P20, and P50. Alexa Flour 568 Dextran dye was continuously puffed within the corpus callosum via a glass pipette. Confocal scans of red fluorescence signal were acquired in grayscale. We then inverted the Look‐Up Table (LUT) to show extracellular spaces as dark regions. (B) To quantify the extracellular space (ECS) width, a line (dark green line) was drawn perpendicular to the axons' direction. From the line profile, the extracellular spaces were identified as negative valleys (blue cross circles marked the peaks). The ECS widths were calculated as the Full Width at Half Maximum (FWHM) of each identified valley, where the maximum amplitude was measured from the baseline (gray dashed line) to the valley peak (blue cross circles). (C) Histogram showing the distribution of all measured ECS widths from three age groups. (D) Violin plots showing ECS width measurements from three age groups. Open circles are individual measurements. The solid horizontal lines denote the mean for each age group. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.05 for overall age group effect. *p < 0.05. N = 1592 measurements from P10 group, N = 1604 measurements form P20 group, N = 2447 measurements from P50 group.

3.4. The Decrease in ECS During Postnatal Development Minimally Affects the Micron‐Scale Small Molecule Diffusion Within the Corpus Callosum

Emerging evidence indicates that myelin development can be influenced by extrinsic factors, such as neuronal activities (Fields 2015; Mount and Monje 2017). In mouse corpus callosum, unmyelinated axons ectopically release glutamate via discrete axonal vesicle release machinery (Kukley et al. 2007; Ziskin et al. 2007). Only a fraction of axonal release sites make synaptic contacts with postsynaptic OPCs, and many others likely directly release glutamate into the ECS (Kukley et al. 2007). In addition to neurotransmitters, various extracellular signaling molecules, growth factors and cytokines may diffuse through the ECS and have been reported to regulate myelin formation (Hill and Nishiyama 2014; Bergles and Richardson 2015). Collectively, these findings suggest that the diffusion kinetics of extracellular molecules within the corpus callosum may significantly impact oligodendrocyte lineage cell behaviors and myelin formation. Our measurement of ECS widths displayed a sharp decline from P10 to P20 in mouse corpus callosum. Here, we sought to answer the question of whether this reduction in extracellular space affects molecule diffusion kinetics.

To quantitively measure the molecule diffusion kinetics, we ejected a red fluorescent dye (Alexa 594, Molecular weight: 758.79) from a tiny glass pipette tip within the corpus callosum with a brief electrical stimulus. A straight line across the dye ejection side was repetitively scanned to capture the time‐lapse fluorescence spread along the drawn line (Figure 5A, top). When we compared the fluorescence spread between line‐scan images obtained parallel and perpendicular to the axon bundle orientation, we were able to investigate the molecule diffusion range and anisotropy within the corpus callosum (Figure 5A). Line‐scan images clearly showed that the fluorescent dye spread laterally overtime (Figure 5A,B). By plotting the peak fluorescence intensity at increasing distances from the dye ejection site, we observed a consistent decay pattern and fitted it with an exponential curve. From this fit, we extracted the length constant (λ), which serves as an indicator of the effective diffusion range (Figure 5C). We then compared the λ across all age groups, both parallel and perpendicular to callosal axon direction. Our results showed that the diffusion range was comparable across all three age groups, and no anisotropy was observed in the dye diffusion pattern between directions parallel and perpendicular to axon bundles (Figure 5D,E).

FIGURE 5.

Fluorescent dye diffuses within the corpus callosum extracellular space isotropically and the diffusion range remains constant during postnatal myelin development. (A) Top cartoon: A glass pipette filled with Alexa594 fluorescent dye was placed in the corpus callosum (light gray shadings indicate callosal axons). Dye was ejected from the pipette tip with a brief electrical stimulus (light blue lightening sign). Confocal line‐scans were performed through the dye ejection site, oriented either perpendicular or parallel to callosal axons (dark dashed lines). Bottom: Example line‐scan image shows the lateral spread of red fluorescence signal along the scanned line (x‐axis) over time (y‐axis). LUT scaling was adjusted to make the distant edges of dye diffusion more visible to the reader. However, no pseudo‐saturation was introduced during LUT adjustment. Blue dashed lines (# 1, 2, 3, and 4) mark the locations where line profiles in B were obtained. Scale bars: 5 μm, 10 ms. (B) Fluorescence line profiles (width = 1 μm, smoothed with a sliding average of five points) obtained from position #1, 2, 3, and 4 in A. Peak fluorescence amplitudes for each line profiles were normalized to the maximum signal at the dye ejection site. Gray dashed lines mark the positions where the crosshair cursors were placed for peak amplitude measurements for each line profile during analysis. Note that as the distance from the ejection site increased from #1 to #4, the peak fluorescence intensity progressively decreased. (C) For each dye ejection, normalized peak fluorescence amplitudes (blue open circles) were plotted against their distances from the dye ejection site. An exponential curve was fitted to the data points to model the dye diffusion kinetics, and the length constant (λ) was extracted to indicate the diffusion range. (D) Bar graphs showing comparison of λ between parallel and perpendicular direction across three age groups. p > 0.05 for all comparisons between groups. Mixed effects models with fixed effects of group and direction, and their interaction for diffusion analysis. Each line scan image generated 2 fluorescence decay pattern data sets. In total, N = 18 for parallel and N = 18 data sets for perpendicular direction in P10 group, N = 10 for parallel and N = 8 data sets for perpendicular direction in P20 group, N = 16 for parallel and N = 20 data sets for perpendicular direction in P50 group. (E) Average fluorescence decay patterns for P10 and P50 groups in both parallel and perpendicular directions. N = 18 for parallel and N = 18 for perpendicular direction in P10 group, N = 16 for parallel and N = 20 for perpendicular direction in P50 group.

3.5. The Expression of Glutamate Transporter, GLT‐1, and the Density of VAMP2+ Vesicles Markedly Increased Within the Corpus Callosum During Postnatal Development

Glutamate transporters actively participate in glutamate signaling by clearing excess extracellular glutamate and maintaining the ambient glutamate concentration within the ECS. We next performed immunoblotting to characterize the expression patterns of two major glutamate transporters, GLT‐1 and GLAST, using mouse corpus callosum tissue from different postnatal developmental time points (Figure 6A). Immunoblotting results demonstrated that the expression of GLT‐1 increased more than two folds from P10 to P20 and maintained at the high expression level at P50 (Figure 6B). The expression of GLAST remained comparable across the three age groups (Figure 6B).

FIGURE 6.

Postnatal myelin development in mouse corpus callosum is accompanied by the upregulation of glutamate transporter, GLT‐1, and GLT‐1 is expressed by both astrocytes and oligodendrocyte lineage cells. (A) Immunoblot shows the expression of glutamate transporters (GLT‐1 and GLAST) at three different time points. (B) Bar graphs show the quantification of immunoblot in A. One‐way ANOVA with Tukey's HSD test for multi‐group comparison. For GLT‐1 comparison, p < 0.05 for overall age group effect. *p < 0.05. For GLAST comparison, p > 0.05 for overall age group effect. N = 3 in each age group. (C) Single focal plane confocal images showing the immunostaining of GFP, GLT‐1 and DAPI signals in corpus callosum of Aldh1l1‐eGFP mice. White arrowheads point to GLT‐1+ and GFP+ astrocytes. White dashed lines outline the corpus callosum regions. (D) Single focal plane confocal images showing the immunostaining of GLT‐1, Olig2 and DAPI in corpus callosum of Aldh1l1‐eGFP mice. White arrowheads point to GLT‐1+ and Olig2+ oligodendrocyte lineage cells. (E) Bar graphs show the density of GLT1+ cell bodies across the three age groups. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p > 0.05 for overall age group effect. (F) Bar graphs demonstrate the density of GFP+ astrocytes within corpus callosum. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p > 0.05 for overall age group effect. (G) Quantification of the percentage of GLT‐1+ cell bodies identified as GFP+ astrocytes. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.001 for overall age group effect. *p < 0.05, **p < 0.01, ***p < 0.001. (H) Quantification of the percentage of GLT‐1+ somata identified as Olig2+ oligodendrocyte lineage cells. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p > 0.05 for overall age group effect. (I) Quantification of the fraction of Olig2+ cells with GLT‐1 immunoreactivity at the cell soma. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p > 0.05 for overall age group effect. (J) Quantification of the fraction of GFP+ astrocytes exhibiting somatic GLT‐1 expression. Mixed effects models to test the group effect on outcomes with Tukey adjustment. p < 0.001 for overall age group effect. **p < 0.01, ***p < 0.001. For (E) to (J), N = 4 mice for P10, N = 4 for P20 and N = 5 for P50 group.

Next, we aimed to identify the cellular sources that express GLT‐1 transporters. Astrocytes are considered the major glutamate transporter cellular source in CNS, particularly in gray matter regions. To visualize astrocytes within the corpus callosum, we utilized a transgenic mouse line that expresses eGFP under the Aldh1l1 promoter control (Yang et al. 2011). Brain sections of Aldh1l1‐eGFP mice from the same three age groups were prepared and immunolabeled with the antibody against GLT‐1 and the nanobody against GFP (Figure 6C). We were able to identify GLT‐1+ cell bodies from the confocal images (Figure 6C). The overall GLT‐1+ cell density was comparable across all three age groups (Figure 6E, p > 0.05 for overall age group effect). The density of GFP+ astrocytes was also comparable across all three age groups (Figure 6F, p > 0.05 for overall age group effect). To our surprise, we noticed that only a fraction of GLT‐1+ cells were GFP+ astrocytes (Figure 6C,G): In the P10 group, 37.78% ± 2.5% GLT‐1+ cells were GFP+ astrocytes. This fraction further decreased to 16.99% ± 1.6% at P20 (Figure 6G, p < 0.001 for P10 vs. P20), and 25.73% ± 1.5% at P50 (Figure 6G, p < 0.01 for P10 vs. P50, p < 0.05 for P20 vs. P50). To identify additional cell types expressing GLT‐1, we performed double‐immunolabeling in the corpus callosum using antibodies against GLT‐1 and oligodendrocyte lineage marker Olig2 (Figure 6D). Our quantification revealed that ~50%–60% of GLT‐1+ cells were also immunoreactive for Olig2 across all three age groups (Figure 6H, p > 0.05 for overall age group effect), suggesting substantial expression of GLT‐1 by oligodendrocyte lineage within corpus callosum white matter tract. However, not all Olig2+ cells express GLT‐1 somatically. Overall, approximately 40%–50% Olig2+ cells were GLT‐1+ (Figure 6I, p > 0.05 for overall age group effect). Similarly, we observed many GFP+ astrocytes that lacked GLT‐1 expression at their soma. At P10, 69.67% ± 4.0% of GFP+ astrocytes co‐express GLT‐1 (Figure 6J). This fraction significantly decreased to 39.58% ± 2.9% at P20 and recovered back to 62.30% ± 4.1% by P50 (Figure 6J, p < 0.001 for P10 vs. P20, p < 0.01 for P20 vs. P50). In summary, our results suggest that the developmental increase in GLT‐1 protein levels is likely driven by the upregulation of GLT‐1 within individual cells rather than an expansion of GLT‐1‐expressing cell population. Furthermore, our data demonstrated that astrocytes are not the sole source of GLT‐1 expression in mouse corpus callosum. Instead, a substantial fraction of oligodendrocyte lineage cells also express GLT‐1.

Finally, we aimed to examine whether the number of axonal vesicle release machinery changes during postnatal myelin development. We first immunostained for VAMP2+ vesicles (Figure 7A) and quantified the density of VAMP2+ puncta within the corpus callosum. Our results revealed an increase in VAMP2+ puncta density from P10 to P50 (Figure 7B, p < 0.01 for overall age group effect. p < 0.01 for P10 vs. P50). To further determine whether the increase in VAMP2+ vesicle density resulted from a higher frequency of release sites along callosal axons, we sparsely labeled callosal axons with tdTomato fluorescence protein by transducing cortical neurons projecting through corpus callosum with AAV‐tdTomato virus. We then quantified VAMP2+ puncta along tdTomato+ callosal axons (Figure 7C). Due to the limited thickness of our thin frozen sections, the traced axons display various lengths. Nevertheless, the overall length distributions were comparable across the three age groups (Figure 7D). Our analysis demonstrated a higher percentage of traced axons lacking VAMP2+ puncta in the P10 group (~70%) compared to the P20 and P50 groups (~55%, Figure 7E, p < 0.01 for overall age group effect. p < 0.01 for P10 vs. P20, p < 0.01 for P10 vs. P50). However, from those traced axons co‐localized with VAMP2+ puncta, the cumulative probability plot of VAMP2+ puncta frequency was comparable across the three age groups (Figure 7F). Our results suggest that the ectopic axonal vesicle release machinery continues the maturation process during early postnatal development.

4. Discussion

In this study, we demonstrated that OPC density peaks early postnatally but declines after the third postnatal week in mouse corpus callosum, consistent with the general myelin development timeline within CNS (Nishiyama et al. 2021). This decline in OPC density is accompanied by a progressive increase in oligodendrocyte density from P10 to P50 (Figure 2). Programmed cell death of developmental oligodendrocytes, particularly pre‐myelinating oligodendrocytes, has been reported in the optic nerve, spinal cord and cerebral cortex, with estimated death rates ranging from ~20% to 50% (Barres et al. 1992; Trapp et al. 1997; Calver et al. 1998). In our study, we demonstrated that 30%–40% of CC1+ oligodendrocytes within corpus callosum express pro‐Caspase‐3 during the first 3 postnatal weeks. However, despite turning on the Caspase‐3 machinery, these cells do not appear to undergo programmed cell death since they did not show immunoreactivity for the active form of the enzyme, cleaved Caspase‐3, that is required for active apoptosis (Figure 2). The fraction of pro‐Caspase‐3+ oligodendrocytes declined to bellow 20% between P20 and P50 (Figure 2). Our results suggest a transient upregulation of pro‐Caspase‐3 in differentiated oligodendrocyte that does not necessarily lead to the cell death fate. Interestingly, a recent study (Kamen et al. 2025) reported similar results as our study that pro‐Caspase‐3 are temporally upregulated in differentiating oligodendrocytes in developmental cortex and corpus callosum. Majority of these pro‐Caspase‐3+ cells are not positive for cleaved Caspase‐3 (Kamen et al. 2025). Kamen et al. proposed that the upregulation of pro‐Caspase‐3 enable newly differentiated oligodendrocyte to make the fate decision between programmed cell death and continued differentiation (Kamen et al. 2025). Although generally considered as a mature oligodendrocyte marker, it has been reported that BCAS1+ differentiating oligodendrocytes can also express CC1 (Fard et al. 2017). We postulate that in P10 and P20 group, a large portion of CC1+ cells within corpus callosum are differentiating or newly differentiated oligodendrocytes, while more CC1+ cells in the P50 group are matured oligodendrocytes. While previous studies suggest that developmental oligodendrocyte death is driven by a lack of survival cues or failure to engage in axonal ensheathment (Barres et al. 1992; Trapp et al. 1997), it remains unclear whether axonal activity can bias this fate decision and during myelin development. In addition, it is still unknown whether pro‐Caspase‐3 contributes to any non‐apoptotic functions during oligodendrocyte differentiation.

The formation of myelin sheaths by oligodendrocytes drastically increases the conduction velocity along axons. For unmyelinated axons, the conduction velocity is generally proportional to the square root of the axonal diameter (Hodgkin 1954). Additionally, the expression of ion channels, especially Na⁺ channels, can impact the capability for unmyelinated axons to regenerate action potentials repeatedly. Any alteration in membrane properties, for example, membrane capacitance, might also impact conduction velocity. Electron microscopic studies reported a developmental increase in axon diameter for unmyelinated callosal axons and an overall developmental decrease in total axon counts (Berbel and Innocenti 1988; Kim and Juraska 1997), suggesting that callosal axons undergo projection refinement during postnatal development. Callosal refinement might lead to changes in conduction of unmyelinated axons. For instance, larger axon diameter can result in faster conduction velocity and the shifts in axonal population composition may also impact overall conduction. However, the impact of these factors on axonal conduction is likely very minor compared to the conduction acceleration caused by myelination.

For myelinated axons, action potentials are propagated in a saltatory manner, meaning that they are only actively regenerated at nodes of Ranvier, while passively spreading under the myelin sheaths between nodes. Conduction velocity along myelinated axons is regulated by several factors, including the length of myelin sheath segments, the length of the nodes of Ranvier, and ion channel density at nodes. CAP recordings in adult corpus callosum often display two peaks, wherein the fast peak is considered the CAP carried by myelinated axons while the slower peak is considered the CAP carried by unmyelinated axons (Crawford et al. 2009; Colley et al. 2010). After toxin‐induced focal demyelinating injury, the fast myelinated peak disappears from the CAP recording while the slower unmyelinated peak remains (Sahel et al. 2015). However, it has not been investigated how CAP recording evolves during postnatal myelin development in mouse corpus callosum. Our Fluoromyelin staining and SCoRe imaging data showed that callosal axons progressively increase myelination from P10 to P50 (Figure 1). Consistent with these structural changes, CAP recordings demonstrated corresponding functional shift. We observed that about half of CAP recordings at 1000 μm from P20 group displayed two distinct CAP peaks, whereas ~90% of recordings in P10 groups only exhibited a single peak, which is likely carried by unmyelinated axons (Figure 3). By P50, the majority (75%) of CAP recordings displayed more than one peak (Figure 3). The emergence of multiple CAP peaks was also accompanied by shorter latencies for the first peak in P20 and P50 groups (Figure 3). These data support the idea that progressive myelin development likely contributes to increased axonal conduction by facilitating the transition to saltatory conduction in an increasing number of axons.

When measuring the conduction velocity for the first CAP peak, the velocity at P50 is significantly faster, whereas the P20 group showed measurements comparable to those in the P10 group. Considering that ~half of P20 CAP recordings did not exhibit a noticeably faster separated CAP peak (Figure 3D), it is possible that the myelination level at P20 has not yet reached a level sufficient to significantly enhance axonal conduction velocity. Several factors might contribute to this phenomenon. For example, the myelin sheaths formed at P20 might be shorter. A time‐lapse imaging study of zebrafish spinal cord myelin development reported that initial ensheathment only formed short myelin segments, and that neuronal activity stabilized and elongated myelin sheaths along active axons (Hines et al. 2015). Longer myelin sheaths allow electrical impulses to travel passively over longer distance beneath the sheath, thereby increasing conduction velocity. It is plausible that myelin sheaths at P20 are generally shorter than P50. Additionally, saltatory conduction relies on the asymmetric distribution of ion channels along myelinated axons: high density of voltage‐gated Na⁺ channels are clustered at the nodes of Ranvier, where they are responsible for regenerating action potentials, while Shaker‐type voltage‐gated K⁺ channels are clustered at juxtaparanode region beneath the myelin sheath (Peles and Salzer 2000; Poliak and Peles 2003). These K⁺ channels contribute to maintaining the resting membrane potential in internode regions (Peles and Salzer 2000). It is possible that the nodal structure and ion channel distribution are not yet fully matured around P20, thereby limiting the conduction velocity. Future studies characterizing myelin coverage, sheath and nodal lengths, and ion channel distribution along callosal axons will provide further insights into the evolution and optimization of axonal conduction in the corpus callosum during myelin development.

Within the corpus callosum, individual axons are separated by narrow ECS that contains a network of extracellular matrix (ECM) molecules and interstitial fluid. This local ECS microenvironment is crucial for oligodendrogenesis and myelin formation, because the interstitial fluid serves as a reservoir for various ions, neurotransmitters, growth factors and cytokines, all of which can impact myelin development (Hill and Nishiyama 2014; Bergles and Richardson 2015). The migration, survival and differentiation of OPCs are also supported by various ECM molecules (Milner et al. 1996; Yamada et al. 2022). However, traditional experimental approaches for studying ECS are limited and cannot directly visualize ECS with a decent resolution in live tissue (Hrabetova et al. 2018). Recently, Super‐Resolution Shadow Imaging (SUSHI) was developed to directly image the extracellular fluid using STED microscopy (Tonnesen et al. 2018). The principle of this technique is to label the extracellular fluid with a fluorescent dye and acquire the fluorescent ECS images where the cellular structures are dark. During imaging processing, these images are inverted to display ECS as dark regions. Notably, this shadow imaging technique is not limited to super‐resolution microscopes but is also applicable for confocal and 2‐photon microscopy systems (Dembitskaya et al. 2023). Here, we adopted this shadow imaging technique using a laser‐scanning confocal microscope to visualize ECS at different myelin developmental stages within the corpus callosum using acute brain slices (Figure 4). We observed the largest ECS width between callosal axons at the early stage of myelin development around P10, and ECS width sharply decreased when myelination progressed by around P20. Our quantification of ECS width did not show difference between P20 and P50 groups (Figure 4). However, given the excitation laser wavelength of 561 nm and the objective numerical aperture (NA) of 1.0, the theoretical lateral (x‐y axis) resolution limit of our system is ~220 nm. We recognize that any subtle difference in ECS widths between P20 and P50 may fall below our optical system's resolution limit. Nevertheless, the quantification (Figure 4D) clearly demonstrates that the drop in ECS width from P10 to P20 represents a more substantial change.

The diffusion coefficient (D) of a single molecule in a 3‐dimensional liquid generally follows the Stokes–Einstein equation:

$$D=\frac{\mathit{\kappa T}}{6\mathit{\pi \eta r}},$$

where k is Boltzmann's constant, T is the absolute temperature, η is the dynamic viscosity of the liquid, and r is the hydrodynamic radius. If we assume the geometry of the molecule is spherical, where the molecule volume scales with the mass, then D will be inversely proportional to the cube root of the molecular weight ($\sqrt[3]{\mathrm{MW}}$). In the extracellular space in the brain, the diffusion of molecules is partially restricted by the cellular structural obstacles, which is quantified by tortuosity, λ (Nicholson and Phillips 1981). The effective diffusion coefficient D*, is determined by D/λ ² (Nicholson and Tao 1993). Previous studies reported that for ions with very small MWs, such as Na⁺, D* is ~590 μm²/s (Nicholson 2001), while the D* for glutamate (MW ~146) is approximately 300 μm²/s (Nielsen et al. 2004; Zheng et al. 2008). For molecules with much larger MWs (e.g., 40–70 kDa), the D* decreases to the range of 7–9 μm²/s (Nicholson and Tao 1993), likely due to the physical constraints of small gaps within the brain tissue. In this study, we used Alexa 594 with a MW of ~759 to probe the diffusion in corpus callosum during postnatal development. Although its MW is larger than that of glutamate, it remains within the same order of magnitude as glutamate and several other important neuromodulators, such as adenosine (MW ~267) and ATP (MW ~507). Based on the inverse relationship between free diffusion coefficient D and $\sqrt[3]{\mathrm{MW}}$, the theoretical D for Alexa594 is only slightly smaller than that of glutamate (approximately 1.7‐fold). In our data, the length constant λ for dye decay across age groups and scanning directions is between 3 and 4 μm (Figure 5D). These values align well with reported action range of glutamate in the brain. In fact, it has been recently reported that glutamate action range in mouse hippocampus for high‐affinity NMDA receptors is approximately 1.5 μm and the glutamate action range for low‐affinity AMPA receptors is approximately 450 nm (Matthews et al. 2022). Given that the action range of glutamate is significantly restricted by glutamate transporter activities, the pure diffusion range of glutamate is expected to be very close to our estimate using Alexa 594. Therefore, our experiments provide a reliable estimate of micron‐scale diffusion for small molecule such as glutamate and other signaling molecules with relatively small MWs. However, we acknowledge that the diffusion dynamics will be different for much larger molecules, such as growth factors.

During postnatal callosal development, many alterations may occur to impact effective diffusion coefficient for signaling molecules, including the density and viscosity of extracellular liquid, and the geometry of the ECS. Alteration in ECM network might also impact molecule diffusion dynamics. In this study, we were surprised to find no significant difference at micron scale in diffusion range across three developmental stages (Figure 5), despite the sharp reduction of ECS width from P10 to P20 as mentioned above. One possible explanation is that the reduction in ECS between axons does not reach a threshold to generate additional physical hinderance for small molecules such as neurotransmitters and neuromodulators. However, it is possible that larger macromolecules, such as growth factors or other signaling proteins, would exhibit age‐dependent diffusion dynamics. A more detailed biophysical analysis is required to decipher how these microenvironment variables affect diffusion in corpus callosum during myelin development. Additionally, dye diffusion analysis did not reveal any apparent anisotropy at micron scale, despite the uniformed orientation of callosal axon bundle. A recent study analyzing the spread of glutamate in mouse hippocampus reported similar findings: the spread of glutamate within CA1 stratum radiatum was isotropic, despite the uniform orientation of Schaffer collateral axons in this region, and was independent of the amount of glutamate released (Matthews et al. 2022).

Glutamate signaling plays a critical role in regulating myelin development and repair in the context of activity‐dependent myelination (Micu et al. 2006; Lundgaard et al. 2013; Gautier et al. 2015; Kougioumtzidou et al. 2017; Chen et al. 2018; Evonuk et al. 2020; Khawaja et al. 2021). Emerging evidence indicates that callosal axons ectopically release glutamate through discrete axonal vesicle release machinery (Kukley et al. 2007; Ziskin et al. 2007). While some of the vesicle release sites form direct synapses with OPC processes, many do not exhibit clear contact with OPCs (Kukley et al. 2007), suggesting that glutamate is likely released into ECS at these release sites and can potentially activate glutamate receptors expressed by oligodendrocyte lineage cells at a distance. Excess extracellular glutamate is typically cleared by astrocytic glutamate transporters. While perisynaptic glutamate uptake by astrocytes is well characterized in gray matter, much less is known about glutamate transporter expression in white matter. Our immunoblotting data showed that GLT‐1 expression was significantly upregulated from P10 to P20 and maintained at a steady level between P20 and P50 (Figure 6A,B). Immunostaining and cell counting analysis revealed that the density of GLT1+ cells remains comparable from P10 to P50, suggesting that the upregulation of GLT‐1 protein level is not caused by increased number of GLT‐1 expressing cells, but an upregulated GLT‐1 level within individual cells. Accumulating data has demonstrated the expression of glutamate transporters in oligodendrocytes (Suarez‐Pozos et al. 2020). Particularly, DeSilva et al. (Desilva et al. 2007, 2009) reported GLT‐1 expression by oligodendrocytes in developing human brain white matter and rat corpus callosum. Our analysis demonstrated that more than 50% of GLT‐1+ cells within corpus callosum across P10 to P50 groups belong to oligodendrocyte lineage (Figure 6H). In contrast, a smaller fraction (40%) of GLT‐1+ cells are GFP+ astrocytes at P10, and this fraction further decreases at P20 and P50 (Figure 6G). Our data raises the question whether the glutamate uptake function is primarily carried by oligodendrocyte lineage within the corpus callosum. However, we acknowledge that our histological analysis primarily focused on somatic GLT‐1 immunoreactivity, while the majority of glutamate uptake typically occurs in fine processes, and that somatic transporter expression may not necessarily reflect the functional uptake capacity at the process level. Therefore, future electrophysiological studies investigating developmental alterations in glutamate uptake kinetics, particularly in distal processes of both the astrocytes and oligodendrocyte lineage cells, would benefit our understanding of glutamate clearance mechanisms within the corpus callosum. Furthermore, we also discovered substantial numbers of Olig2+ cells and GFP+ astrocytes did not show immunoreactivity for GLT‐1. Whether these GLT‐1 negative cells express other types of glutamate transporters needs to be further investigated.

On the other hand, our quantification of VAMP2+ puncta suggests that at early postnatal time points, for example, P10, the number of available axonal vesicle release machinery is rather low, and the number of axonal machinery increases as myelination progresses (Figure 7). This observation was supported by several pervious electrophysiology studies: Vana et al. recently demonstrated that late embryonic OPCs display little to no synaptic currents and the frequency of miniature post‐synaptic currents gradually increases postnatally from P7 to P12 (Vana et al. 2023); De Biase et al. also reported that hypertonic solution (by including high concentration of sucrose in ACSF) evoked more than 300 miniature synaptic events in P40 OPCs but only evoked less than 50 events in P6 OPCs within mouse corpus callosum (De Biase et al. 2010). Our data demonstrates that the low density of VAMP2+ puncta at P10 is due to fewer axons equipped with vesicle release machinery, not a lower frequency of ectopic release sites along individual axons. Among axons with VAMP2+ puncta, the puncta frequency is comparable across all three age groups (Figure 7). Our data suggests that the structural maturation of axonal vesicle release machinery continues beyond the third postnatal week.

In summary, our study provided a developmental profile of myelin development and oligodendrocyte lineage dynamics within the corpus callosum during postnatal development. The progressive myelination at the structural level aligns closely with the gradual facilitation of callosal axonal conduction. Although glutamate signaling is thought to play a critical role in myelin plasticity within white matter, our findings suggest that the glutamate signaling network continues to refine into later developmental stages. While the diffusion rate of glutamate at micron scale likely remains unaltered across different developmental stages in the corpus callosum, both glutamate transporter expression and axonal vesicle release machinery appear to reach peak levels at later time points. Future studies should explore the functional maturation of synaptic signaling between callosal axons and OPCs, as well as extrasynaptic glutamate signaling within the corpus callosum during postnatal development.

Author Contributions

W.S. conceived the project and designed the research study. H.J., M.B., S.L. and W.S. performed the research and analyzed the data. J.P. performed statistical analysis. W.S. supervised the research and drafted the paper. All authors contributed to editing the manuscript and approved the submitted version.

Funding

This study was supported by National Institute of Neurological Disorders and Stroke (NINDS) R01NS124714 (to W.S.); the startup fund from The Ohio State University College of Medicine (to W.S.).

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Individual data points or distribution of all data points are provided in the data figures. Requests for any other data files should be directed to the corresponding author.