The Influence of Changes in Microglia Development on the Plasticity of the Developing Visual Cortex Circuit in Juvenile Mice

Abstract

Purpose: To investigate the role of microglial subtypes in mouse visual cortex development, focusing on ocular dominance plasticity and interactions with GABAergic neurons and the extracellular matrix.

Methods: Immunofluorescence and single-nucleus RNA-sequencing (snRNA-seq) were used to study microglia in the binocular primary visual cortex (V1) from postnatal day (P) 11 to P42. Gene ontology (GO) analysis assessed synapse organization, and the impact of microglial disruption on ocular dominance plasticity was examined. Visual evoked potentials and miniature postsynaptic current recordings are used to monitor functional changes in V1.

Results: Microglia underwent a marked expansion between P11 and P21 and stabilized after P35, coinciding with notable changes in gene expression that aligned with synaptic remodeling. GO analysis at P14 and P28 revealed significant enrichment in synaptic organization linked to microglia. Single-nucleus RNA sequencing identified six distinct microglial clusters, among which two functionally relevant subpopulations were closely linked to cortical synaptic plasticity. One cluster, enriched in inflammatory responses and endocytosis, peaked at P21, whereas another cluster, associated with synapse organization and signaling, exhibited dynamic changes after eye opening and during the critical period, significantly influencing cortical synaptic plasticity. In parallel, perineuronal nets (PNNs) and PV(+) interneuron populations increased and reached steady levels by P42, suggesting that microglia help coordinate the timing of inhibitory circuit maturation. Disrupting microglial function during the critical period impaired ocular dominance plasticity, but this effect was reversed after treatment cessation. Mechanistically, microglial depletion enhanced PV(+) interneuron numbers, elevated PNN expression, and altered synapse development.

Conclusions: Our findings highlight specific microglial subtypes as key regulators of cortical synapse development and plasticity through their interactions with PV(+) interneurons and PNNs. These insights advance our understanding of microglial contributions to visual cortex development and provide potential avenues for targeting microglial function to modulate cortical plasticity.

Neuronal plasticity refers to structural and functional changes in neuronal circuits in response to experiences. Across multiple sensory systems, especially visual systems, experiences guide the functional maturation of neuronal circuits during periods of heightened plasticity known as “critical periods.” Abnormal experiences during these windows during youth irreversibly alter neuronal connections and functions into adulthood. Cortical circuit function requires balanced production and pruning of synapses, proper timing of synaptic maturation, and activity-dependent structural and functional plasticity of synaptic connections. Thus cell types that regulate synapse structure may play significant roles in the refinement of the functional properties of sensory circuits. 

As resident immune cells and key players in brain pathologies, microglia are increasingly known to play essential roles in regulating neuronal activity, synaptic plasticity, and the normal function of the adult brain. One essential function of microglia in brain development is synaptic pruning, an essential process that contributes to synaptic connectivity and circuit refinement via C1q, C3, and CR3 complement signaling or CD47 and signal-regulated protein alpha signaling pathways.^1–4 By mediating synaptic pruning, resting microglia play an important role in regulating experience-dependent plasticity in the visual cortex after whisker removal or monocular deprivation (MD), as well as in the regulation of learning and memory. Impaired synaptic pruning is associated with weaker synaptic transmission, disrupted regional brain connectivity, deficits in social interaction, and increased repetitive behaviors, which are characteristics of autism spectrum disorders. Further corroborating the role of microglia in synaptic pruning, deficient microglial autophagy impairs synaptic pruning, with long-lasting repercussions resulting in social behavioral defects. In addition, microglial activation or dysfunction has been associated with the progression of cognitive deficits in normal aging, neurodegenerative and neurological diseases such as Alzheimer's disease, traumatic brain injury, and HIV-related neurocognitive disorders.^5–9 

During critical periods of development, microglia respond to changes in sensory input and promote salient remodeling. The destruction of microglia during the critical period can significantly affect neural development.^1,4,10 Experience-dependent pruning of synapses in the visual cortex is evident at four weeks postnatal, when microglia exhibit a mature ramified phenotype.¹¹ Microglial contact-induced filopodium formation in mice is observed only at P8–P10, corresponding to when microglia are in an amoeboid, activated state. Compared with those in the somatosensory cortex of P12–P14 mice and adult mice, microglia in the L2/3 somatosensory cortex of P8–P10 mice exhibit a more activated phenotype, as confirmed by morphological analysis.^4,12 The current consensus regarding microglial homeostasis in adulthood is that there is limited spatial heterogeneity of microglia across anatomically distinct central nervous system (CNS) regions.¹³ The dichotomy of “good” versus “bad” microglia, such as resting versus activated and M1 versus M2 microglia, is inconsistent with the diverse repertoire of microglial states and functions that have been elucidated in recent years in relation to development, plasticity, aging, and disease.¹⁴ Microglia have a complex “sensome,” a series of surface receptors that allow them to detect changes in their environment. Therefore microglia exist in diverse, dynamic, and multidimensional states depending on context. The key variables driving the phenotypic transformation of microglia are distinct signaling pathways regulated at multiple levels (e.g., transcriptional, epigenetic, translational, and metabolic levels), and each pathway governs distinct microglial functions or properties.¹⁵ Microglia display an activated morphology and have high phagocytic activity during the postnatal period (the first three weeks after birth).¹⁶ Microglia proliferate within the first two weeks postnatally, and their number in the CNS declines by 50% between the third and sixth weeks postnatally. This period of microglial proliferation is consistent with CPs, which demonstrates the important role of microglia in visual and somatosensory experience-dependent plasticity.^17,18 Colony stimulating factor receptor 1 (CSFR1), a tyrosine kinase receptor, is required for microglia to survive and proliferate.^19,20 The inhibitory effect of CSF1R inhibitors results in the elimination of virtually all microglia from the adult CNS with no adverse effects or deficits in behavior or cognition. The administration of selective CSF1R inhibitors that cross the blood–brain barrier rapidly eliminates 95% of all microglia in the CNS and suppresses microglial function for as long as treatment is maintained. Withdrawal of the inhibitor results in the rapid regeneration of new cells, which then differentiate into microglia.²¹ 

The extracellular matrix (ECM), specifically the condensed ECM structures known as perineuronal nets (PNNs), is an essential component of synapses involved in the regulation of plasticity.¹² PNNs appear during development around the time that the critical periods for developmental plasticity end and are important for both the onset and termination of these critical periods. PNNs preferentially form around fast-spiking parvalbumin (PV)(+) GABAergic interneurons throughout the brain during the termination of the critical period of plasticity, effectively “locking” proximal synapses in place and providing synaptic stability.²² Chronic microglial depletion in a mouse model of Alzheimer's disease rescues dendritic spine loss and prevents neuronal loss in the brain as well as the extensive loss of PNNs. Conversely, enzymatic digestion of CSPGs (the key components of PNNs) in the adult mouse visual cortex has been found to alter the balance between inhibitory and excitatory spiking by reducing inhibitory activity, causing the network to revert to an immature juvenile state.²³ Thus it is important to further understand whether and how microglia regulate neuronal connections in normal and diseased young brains. There are still uncertainties regarding the developmental alterations in different types of microglia and how they influence the plasticity of the primary visual cortex (V1). 

In this study, we investigated the developmental changes in microglia in V1 using single-nucleus RNA sequencing (snRNA-seq) reanalysis. We monitored the development of microglia, PV(+) interneurons, and PNNs. To further understand their roles in development, we employed established chemical methods to deplete the majority of microglia during the critical period of visual cortex development. Our findings indicate that microglia, particularly cluster_0 and cluster_1 are essential for plasticity during the critical period, closely interacting with PV(+) interneurons and contributing to the remodeling of PNNs in early visual circuit development. Moreover, our results elucidate a cellular mechanism that underpins the significant impact of microglia on developmental outcomes related to cortical circuits. 

Methods

Animals

All experiments were performed on C57BL/6J mice (of both sexes) obtained from Beijing Huafukang Biotechnology Co., Ltd. (Beijing, China). The animals were kept in standard cages under standard conditions (temperature of 22°C, 12-hour light/dark cycle) with food and water provided as desired. The number of male and female mice in each group was kept equal. Some of the mice were treated with the CSF1R inhibitor PLX3397 (rodent chow containing 290 ppm PLX3397; Moldiets; BioPike, Shanghai, China) to eliminate microglia. For comparison of the effect of MD, a cohort of mice of the same age was used as the control group. All procedures were reviewed and approved by the Animal Ethics and Welfare Committee (approval No. IRM-DWLL-2020179). 

MD Procedure

The mice were anesthetized with sodium pentobarbital (80 mg/kg, ip). The edges of the upper and lower eyelids were disinfected with betadine, and then the edges of the eyelids were cut off. The surgical eye (right eye) was closed with a 10-0 nylon suture, and tobramycin and dexamethasone eye ointment were applied. After the surgery was completed, the animals were placed on a thermostatic blanket for recovery from anesthesia. The wound was monitored every day, and the animal was excluded from the experiment if the sutures were discontinuous or if the wound became infected. 

Pattern Visual Evoked Potential Recording and Visual Acuity

The mice were anesthetized with sodium pentobarbital and underwent surgery the day before recording. We implanted a homemade platinum-iridium wire electrode in the contralateral (left) V1B area as the recording electrode and in the frontal lobe as the reference electrode. For pattern visual evoked potential recordings, we exposed the animals to sinusoidal gratings at 0.02 cycles per degree (CPD) with 100% contrast and phase reversal at 2 Hz. The display was a computer screen positioned 20 cm directly in front of each eye of the mouse, and each eye was tested in isolation by placing a black occluder in front of the other eye during recording. The eyes were kept open and frequently lubricated with hydrating drops. Body temperature was maintained at 37°C with a heating pad. We used a Roland electrophysiological instrument system for recording and analysis. 

The stimulus for visual acuity recording was a horizontally oriented sine wave grating (mean luminance of 25 cd/m², area of 24 × 26 cm) viewed at a distance of 20 cm. For visual evoked potential (VEP) recording, stimuli were vertical square gratings of different spatial frequencies (0.05, 0.065, 0.084, 0.11, 0.18, 0.23, 0.3, 0.39, 0.5, 0.65, and 0.8 CPD) with 100% contrast, and the temporal frequency was 2 Hz. Custom-made silicone oil contact lenses were applied to the cornea to prevent cataract formation. Acuity was evaluated by linear extrapolation (semilog coordinates) to 0 V of the set of data points close to the noise level as previously described for mice. 

Immunofluorescence and Confocal Microscopy

Whole brains were collected following transcardial perfusion and overnight postfixation with paraformaldehyde (4%). Coronal tissue sections were cut on a vibratome at 100 µm thickness. The sections were processed free-floating at room temperature (RT) in 24-well plates and blocked with 5% BSA. The sections were incubated with primary antibodies (rabbit anti-Iba1 1:200, ab178847 [Abcam, Cambridge, MA, USA]; mouse anti-PV 1:1000, no. 235 [Swant, Bellinzona, Switzerland]; Wisteria floribunda agglutinin (WFA), 1:200, Vector B-1355-2 [Vector Laboratories, Burlingame, CA, USA]; GFAP Mouse mAb, 1:500, no. 3670S [Cell Signaling Technology, Danvers, MA, USA]; VGLUT antibody 1:500, Synaptic System 1-135011 [Synaptic Systems, Goettingen, Germany]) overnight at 4 °C and then incubated with appropriate Alexa-488/647 conjugated secondary antibodies and Streptavidin Alexa Fluor 594 (Abcam) for two hours at RT in the dark. The sections were stained with DAPI (1:1000) for 10 minutes at RT away from light. The fluorescence signals in the brain sections were viewed and imaged under a Leica confocal microscope using a 20 × objective. The data shown in the figures were obtained and are presented in reference to neuronal staining with DAPI and a mouse cortical map. Each data point represents measurements for at least five coronal sections from each animal, and three animals of each postnatal age were used. All confocal images were acquired as Tiff files and analyzed with ImageJ software. The significance of the differences in the data was analyzed by one-way ANOVA. 

Microglial Morphology Quantification

The quantitative analysis of microglial morphology was conducted using ImageJ and its associated plugins, including Analyze Skeleton (2D/3D).^24–26 To systematically evaluate microglial structural changes, three key morphological parameters were quantified: 

Soma Area

The size of the microglial cell body was measured to assess overall enlargement or shrinkage, which may reflect different activation states. 

Endpoints

The number of terminal points of microglial processes was counted as an indicator of cellular ramification and morphological complexity. 

Summed Process Length

The total length of microglial processes was quantified to evaluate the extent of microglial arborization and surveillance capacity. 

High-resolution images of fixed tissue sections stained for Iba1, a well-established microglial marker, underwent preprocessing, including background subtraction, threshold adjustment, and conversion to binary format. The Analyze Skeleton plugin was subsequently applied to extract and quantify endpoints and process length from the skeletonized images, ensuring an objective and reproducible assessment of microglial morphology. 

Analysis of Single-Nucleus Transcriptomics Data

Single-nucleus transcriptomics data for the mouse visual cortex at six postnatal time points (P8, P14, P17, P21, P28, and P38) used in this study were obtained from the NCBI's Gene Expression Omnibus database under the accession number GSE190940 originally published by Cheng et al.²⁷ The cells were subsequently clustered and projected in 2D via uniform manifold approximation and projection (UMAP) with a previously described algorithm. Canonical marker genes for cortical neural and nonneural cells were used to identify the clusters.^27–29 Microglia were identified according to the expression of canonical markers (Cx3cr1, P2ry12, Tmem119, and Ctss) and clustered again. Differentially expressed genes (DEGs) between each of the two groups and specific highly expressed genes (HEGs) of each microglial subtype over time points were performed using “FindMarkers” in the Seurat package (version 4.3.0). DEGs and HEGs with P values < 0.05 were identified. Gene ontology (GO) enrichment analysis of biological process terms was performed with Metascape, and P values < 0.01 were considered to indicate significant enrichment. To determine the degree of communication between microglia and neurons, intercellular communication networks were analyzed and visualized using snRNA-seq data processed with CellChat (https://github.com/sqjin/CellChat). In our analysis, microglia were set as ligand-expressing cells, whereas PV(+) interneurons and excitatory neurons were designated as receptor-expressing cells. Ligand-receptor interactions were predicted using the “netVisual_bubble” function in CellChat. Statistical significance was assessed using permutation testing, where group labels of cells were randomly permuted to generate a null distribution.³⁰ Interactions with a P value < 0.05 were considered statistically significant. 

Electrophysiology

Slice Preparation

Visual cortex slices were prepared as previously described.³¹ The slices of P28 control group or PLX3397 group and immunofluorescence-stained brain slices were taken from the same mice. Briefly, 300 µm cortical slices from C57BL/6 mice of both sexes were cut in ice-cold dissection buffer containing (in mM) 212.7 sucrose, 5 KCl, 1.25 NaH₂PO₄, 10 MgCl₂, 0.5 CaCl₂, 26 NaHCO₃, and 10 dextrose bubbled with 95% O₂/5% CO₂ (pH 7.4). 

The slices were transferred to artificial cerebrospinal fluid and incubated at 30°C for 30 minutes and then at RT for at least 30 minutes before recording. The artificial cerebrospinal fluid was similar to dissection buffer except that the sucrose was replaced with 124 mM NaCl, the MgCl₂ concentration was decreased to 1 mM, and the CaCl₂ concentration was increased to 2 mM. 

Whole-Cell Recording

Whole-cell recordings were made from pyramidal neurons in layers II/III of the monocular V1 region with glass pipettes (3-6 MΩ). Cells with an access resistance <20 MΩ (8–18 MΩ) and an input resistance >100 MΩ were studied. The data were filtered at 2 kHz and digitized at 5-10 kHz using Igor Pro (WaveMetrics Inc., Lake Oswego, OR, USA). Cells were excluded if the input or series resistance changed >20%. 

Miniature Postsynaptic Current Recordings

The following internal solutions were used for recording miniature excitatory postsynaptic currents (mEPSCs) and miniature inhibitory postsynaptic currents (mIPSCs) (in mM): 125 Cs-gluconate, 8 KCl, 1 EGTA, 10 HEPES, 4 (Mg)ATP, 0.5 (Na)GTP, and 5 QX-314 (pH adjusted to 7.25 with CsOH, 280-290 mOsm). Responses were recorded in the presence of 100 mM APV and 1 µM TTX. The reversal potentials for mEPSCSs and mIPSCs were 10 mV and −55 mV, respectively, without compensating for the junction potential. The mEPSCs and mIPSCs were analyzed using the MiniAnalysis program (Synaptosoft, Decatur, GA). Only cells with a root mean square (RMS) noise < 2 (mEPSCs) or < 4 (mIPSCs) were included in the analysis, and the event detection threshold was set at 3 times the RMS noise. Three hundred events with a rise time <3 msec (mEPSCs) or <5 msec (mIPSCs) were selected for each cell to calculate the frequency and amplitude. Nonoverlapping events were used to construct the averaged traces. 

Statistical Analysis

All in vitro electrophysiology data were analyzed with two-tailed t tests or Wilcoxon rank-sum tests, as indicated in the Results section (GraphPad Prism, San Diego, CA). The cumulative distributions were compared with the Kolmogorov‒Smirnov (K-S) test. The sample size is displayed in Figure 7 as the number of cells or the number of animals. The other data were analyzed with two-tailed t tests or Mann‒Whitney U tests for comparisons between two groups and by ANOVA with least significant difference t tests or Tamhane's T2 tests for comparisons between multiple groups. 

All experimental data were subjected to homogeneity of variance tests and normality tests. In cases where the data were normally distributed and assumptions of homogeneity of variance were met, parametric tests were used. If the data were not normally distributed or had heterogeneous variance, nonparametric tests were used. The lines and error bars in all dot plots indicate the means and SEMs. P < 0.05 was considered significant. 

Results

Changes in Microglial Development in the Visual Cortex of Mice

We aimed to describe the timeline of microglial developmental alterations in the mouse visual cortex during postnatal brain development using Iba-1 staining (Fig. 1A). At P8, microglia were sparse, with minimal branching and a rounded morphology (Fig. 1B). The number of Iba1+ microglia significantly increased between P11 and P21 (P < 0.001, one-way ANOVA) and peaked around P21, after which their density gradually declined and stabilized between P42 and P60 (P = 0.75, one-way ANOVA, Fig. 1C). Morphological changes in microglia were also observed throughout postnatal development (Fig. 1B). At P8, microglia exhibited a small soma with few or no processes. From P11 onward, the cells began to extend more elaborate processes, with increased branching complexity and length, indicative of activation. By P28, the soma size decreased, and processes became thinner and longer, suggesting a transition toward a mature, surveillant state (Figs. 1B, 1C). To further quantify these morphological alterations, we analyzed three key parameters: soma area, number of endpoints, and summed process length (Fig. 1C). The soma area significantly decreased from P8 to P28 and remained stable thereafter, consistent with the transition from an activated to a resting state. The number of endpoints increased significantly from P8 to P28 (P < 0.001) and plateaued at later stages, indicating enhanced arborization. Similarly, the summed process length increased progressively from P8 to P28 (P < 0.05) but stabilized at later stages, reflecting the establishment of a mature ramified morphology. 

Figure 1.

 

The microglial cell number and proportion increased within the first two weeks and decreased after the fourth week. (A) Timeline of microglial development in normal mice. (B) Representative images of the V1B region stained for Iba1 in normal mice. (C) The microglial density in the brain expressed as the number of microglia per 1 mm² (N represents the density of microglia in each visual field), area of soma, endpoints and summed process length. (D) Left: UMAP visualization of V1 during postnatal development; right: the cells at each postnatal time point are displayed. (E) Dot plot showing the expression patterns of canonical marker genes. (F) The proportions of nonneural cell subclasses varied with age. (G) The proportions of interneuron subclasses were stable with age.

To delve more deeply into the alterations in microglial activity during development, an analysis was conducted on publicly available single-nucleus transcriptomics data from the mouse visual cortex at six different postnatal time points (P8, P14, P17, P21, P28, and P38); the brain region (V1) and time points corresponding to these data aligned with those studied in our research. After dimensionality reduction, the cell types were identified based on the expression of canonical marker genes (Fig. 1D and 1E); specifically, the cells were identified as including excitatory neurons (Slc17a7), PV(+) interneurons (Gad1, Gad2, Pvalb, and Tac1), Sst(+) interneurons (Gad1, Gad2, Sst, Cpne5, and Th), Vip(+) interneurons (Gad1, Gad2, Vip, and Pax6), Lamp5(+) interneurons (Gad1, Gad2, and Lamp5), Frem1(+) interneurons (Gad1, Gad2, and Frem1), Stac(+) interneurons (Gad1, Gad2, and Stac), microglia (P2ry12, Cx3cr1, Ctss, and Tmem119), astrocytes (Aqp4, Aldoc, and Gfap), oligodendrocytes (ODs; Enpp6, Mog, Olig1, and Bcas1), oligodendrocyte progenitor cells (OPCs; Pdgfra), vascular and leptomeningeal cells (Dcn, Bgn, Aox3, Osr1, Lum, and Col1a1), endothelial smooth muscle cells (Cldn5, Pdgfb, Pecam1, Tek, and Itm2a) and pericytes (Vtn). 

Among the nonneuronal cell subclasses, we found that postnatally, the proportion of ODs gradually increased, while the proportion of OPCs gradually decreased, with the opposite trends being observed with age. The proportion of astrocytes gradually decreased. Interestingly, the proportion of microglia peaked at P17 and then gradually declined (Fig. 1F, Supplementary Fig. S1A). Microglia counts in the single-nuclear sequencing data according to each sample similarly showed that the number peaked at P17 and then declined, reaching a plateau at P28 (Supplementary Fig. S1B, Supplementary Table S1). Both the relative proportion of microglia among all cells and the absolute number of microglia follow similar temporal trends. This consistency suggests that our observations on microglial dynamics during development are robust and reliable. Regarding subclasses of neuronal cells, the relative proportions of excitatory neurons and interneurons were stable over time (Supplementary Fig. S1C). The largest proportion of interneurons were PV(+) interneurons, and the relative proportions of most interneuron subclasses were also stable over time (Fig. 1G, Supplementary Fig. S1D). 

Overall, the number of microglia in the mouse visual cortex began to increase when the eyes opened at approximately P11-P14. It reached its peak at P21 and then gradually decreased, stabilizing after P35. The morphology of microglia changed gradually from an activated state at P11 to a resting state by P28. Additionally, snRNA-seq analysis revealed the same changes in the number of microglia. 

SnRNA-seq Revealed Alterations in Microglia and Their Communication With Neurons During the Development of V1

Compared to P38 (the time point marking the end of the critical period), the number of upregulated DEGs gradually decreases at P8, P14, P17, and P21, while the number of downregulated genes begins to increase at P14, indicating dynamic changes in gene expression patterns at different stages of visual development. Furthermore, at the critical developmental time point P28, the number of upregulated DEGs is the highest compared to P21, suggesting that P28 may represent an important turning point for gene expression changes during visual development (Fig. 2A). We identified 37 genes whose expression levels exhibited a transient increase followed by a decrease across the developmental time points P8, P14, P17, P21, P28, and P38 (Supplementary Fig. S2). The upregulated DEGs after eye opening (P14 vs. P8) were significantly enriched in the positive regulation of synapse organization by microglia according to GO analysis (Fig. 2B). In contrast, the downregulated DEGs from P14 to P21 (P17 vs. P14, P21 vs. P17) were enriched in synapse assembly and axon development. The upregulated DEGs at the peak of the critical period (P28 vs. P21) were significantly enriched in the positive regulation of chemical synaptic transmission and transsynaptic signaling in addition to aerobic respiration and ATP metabolism according to GO analysis. The DEGs at the end of the critical period (P38 vs. P28) were significantly enriched in the negative regulation of glial cell activation and endocytosis. As expected, the expression of the Apoe gene on microglia was significantly upregulated after eye opening (P14). However, the expression of the Cx3cr1 and Grn genes, which are associated with glial cell activation and endocytosis, was significantly downregulated after the peak of the critical period (P38 vs. P28) (Fig. 2C). These results indicate that P14 and P28 may be the key points at which microglia perform critical functions. 

Figure 2.

 

snRNA-seq reanalysis reveals microglial developmental changes in V1. (A) Number of differentially expressed genes (DEGs) in microglia according to pairwise comparisons between groups. (B) Dot plot showing the representative five enriched GO terms for the upregulated (red) and downregulated (blue) DEGs between groups. (C) Violin plot depicting the expression levels of representative DEGs in microglia. (D) Network of putative ligand–receptor pairs showing differential cell–cell interactions across microglia and some neurons. The color of the dot indicates the cell type. The weight of the connecting lines indicates the interaction strength of cell‒cell interaction pairs (thin to thick for low to high). All the arrows point to the receptors. (E) Predicted ligand-receptor interactions between microglia and neurons across developmental time points (P8–P38). The size of each dot corresponds to the statistical significance (P value), computed from a one-sided permutation test. The color gradient reflects the interaction probability. Ligand-presenting cells (microglia) are labeled in red, whereas receptor-expressing cells (neurons) are in blue.

To gain further insight into the impact of microglia on neurons during visual cortex development, we examined the interactions between microglia and different types of neurons, including excitatory and inhibitory subtypes (Fig. 2D). Our analysis suggests that the interaction between microglia and PV(+) interneurons remains relatively stronger than with other interneurons across all examined time points. However, there was an indication of a relative decline in interaction strength between microglia and PV(+) interneurons from P28 to P38 (Fig. 2D). These findings imply that microglia may interact selectively with different types of neurons during specific postnatal stages, potentially contributing to the maturation of the visual cortex. Further studies are needed to confirm these interactions and to explore their mechanistic roles in visual cortex development. 

It is well established that the development of inhibitory neurons marks the boundaries of the critical period in visual development.^32–34 Among these inhibitory neurons, PV(+) interneurons are the most abundant. To investigate how microglia may regulate the plasticity of PV(+) interneurons during the development of the visual cortex, we analyzed the signaling pathways involved in the interaction between microglia (as ligands) and PV(+) interneurons (as receptors). The ligand-receptor interactions between microglia and excitatory neurons began to increase at P14, with Pros1-Tyro3, Grn-Sort1, F11r-Jam3, and Entpd1-Adora1 exhibiting the highest interaction scores at P28 (Fig. 2E, Supplementary Fig. S1A). Additionally, in the interactions between microglia and inhibitory neurons, apart from Ptprm-Ptprm and Ptn-Alk, Grn-Sort1 also showed a significant predicted interaction probability at P28 (Fig. 2E, Supplementary Fig. S1A). These findings suggest that microglia may play a crucial role in the development of the visual cortex at the key developmental time points of P14 and P28. Together, these findings indicate that during the development of V1, the functions of microglia are altered, and the interactions between microglia and neurons vary. 

Identification of the Functions Of Different Microglial Subtypes During Postnatal Development Through snRNA-seq

To further elucidate the involvement of microglia in the critical phase of visual development, individual microglia were isolated and organized into six distinct clusters (microglia 0-microglia 5, Fig. 3A). Specific highly expressed genes (HEGs) were detected in each cluster of microglia (Fig. 3B). The largest number of HEGs was observed between clusters 1 and 3 (Fig. 3C and Supplementary Fig. S4A). Analysis of the proportion of each cluster across the six postnatal time points showed that among the six microglial clusters, cluster 0 was the most abundant, and cluster 1 showed the most variation in proportion. The proportion of cluster 1 increased significantly after eye opening (P14/P17), decreased significantly after the onset of the critical period and rebounded during the peak period (Fig. 3D). We calculated the HEGs for microglia clusters 0, 1, 2, 3, 4, and 5 (Supplementary Fig. S5), and performed GO analysis of the HEGs in each microglial cluster (Fig. 3E). In microglial cluster 0, we observed high expression of Mertk (Mer tyrosine kinase, which mediates microglial phagocytosis),^35,36 Hexb (a key regulator of microglial homeostasis and synaptic pruning),³⁷ and Tgfbr1 (involved in neuroprotection and synaptic stability regulation).^38,39 In contrast, the microglial cluster 1 exhibited elevated expression of Nrg1 (associated with synapse organization)⁴⁰ and Ptprm (implicated in synapse formation)⁴¹ (Supplementary Fig. S5). We also performed GO analysis of the HEGs for each cluster of microglia at each time point, yielding similar results (Supplementary Fig. S4B). According to GO analysis, the HEGs in clusters 1 and 3 were significantly enriched in synapse organization and signaling over the time points. These findings imply that microglial clusters 1 and 3 may contribute to cortical synaptic plasticity. The HEGs in cluster 5 were significantly enriched in the mitotic cell cycle at P8 and P14, which are the stages when the number of microglia increases the most (Supplementary Fig. S4B). 

Figure 3.

 

Dynamic roles of microglial subtypes in regulating postnatal visual cortex plasticity. (A) Left: UMAP visualization of microglia in V1 during postnatal development; right: microglia at each postnatal time point are displayed. (B) Dot plot showing the expression patterns of genes associated with each microglial subtype. (C) The number of HEGs in each microglial subtype. (D) The proportions of microglial subtypes varied with age. (E) Dot plot showing the representative five enriched GO terms for the HEGs in each microglial subtype. (F) Predicted ligand-receptor interactions between microglia_0/1and PV(+) interneuron across developmental time points (P8–P38). The size of each dot corresponds to the statistical significance (P value). The color gradient reflects the interaction probability. Ligand-presenting cells (microglia_0 and microglia_1) are labeled in red, whereas receptor-expressing cells (PV(+) interneuron) are in blue.

Microglia_0, the cluster with the largest proportion, was enriched in inflammation-related functions and endocytosis (Figs. 3D, 3E). However, compared to microglia_0, microglia_1 may interact more strongly with PV(+) interneurons (Fig. 3F). The interaction between microglia_0 and PV(+) interneuron, as well as microglia_1 and PV(+) interneuron, through the ligand-receptor pair Ptprm-Ptprm began to increase at P28, suggesting an active role of microglia in synapse formation.⁴¹ The interaction between microglia_0 and PV(+) interneuron through pleiotrophin (Ptn)-Alk was enhanced at P14, whereas microglia_1 maintained a higher level of interaction through Ptn-Alk with PV(+) interneuron after P14, indicating that microglia_1 may promote microglial proliferation and neurotrophin secretion by activating the PTN signaling pathway,⁴² while also supporting the polarization of developing neurons.⁴³ The Nrg3-Erbb4 ligand-receptor pair showed similar patterns in both microglia_0-PV(+) interneuron and microglia_1-PV(+) interneuron interactions, with an upregulation at P14 followed by a gradual decrease, which may indicate a role of both microglia_0 and microglia_1 in promoting the formation of excitatory synapses in interneurons.⁴⁴ Microglia_0 did not appear to interact with PV(+) interneuron through Nrg1-Erbb4, whereas microglia_1 exhibited an increasing interaction with Nrg1-Erbb4 after P14, suggesting that microglia_1 may play a more significant role than microglia_0 in the regulation of visual cortex plasticity.^34,45 The interaction between microglia_0 and PV(+) interneuron via Grn-Sort1 appeared stronger than that of microglia_1 and began at P14, which may imply that microglia_0 plays an active role in microglial recruitment and activation⁴⁶ (Fig. 3F, Supplementary Figs. S3A, S3B). The predicted interactions between microglia_0/1 and excitatory neurons suggest that microglia_1 may interact more strongly with excitatory neurons compared to microglia_0 (Supplementary Fig. S6, Supplementary Fig. S3A–C). However, the interaction patterns between microglia_0/1 and excitatory neurons differ from those with PV(+) interneuron, indicating that microglia_0 and microglia_1 may interact with excitatory neurons through distinct ligand-receptor pairs compared to PV(+) interneuron. The analysis results above indicate that among the microglial subtypes we defined, clusters 0 and 1 account for the largest proportion of microglia and may play a major role in intercellular interactions with PV(+) interneurons. 

Experience-Dependent Changes in the Development of PNNs and PV(+) Interneurons in the Mouse Visual Cortex

To describe the developmental changes in PV(+) interneurons and PNN formation in the mouse visual cortex, we used WFA to label PNNs (Fig. 4A). We found that few cortical cells were stained by WFA in the mouse visual cortex at P8, and only a small number of cortical neurons were wrapped by PNNs at P14. The density of PNN-wrapped neurons reached a plateau at approximately P42 to P60 and did not decrease further (p = 0.75, one-way ANOVA). A significant increase in the density of PNN-wrapped neurons was found between P14 and P28 (Fig. 4B, P < 0.001, one-way ANOVA), although there was also a relatively smaller increase in the PNN-wrapped cell density between P28 and P42 (Fig. 4B, P < 0.001, one-way ANOVA). The density of PNN-wrapped neurons reached a plateau between P42 and P60 and did not increase further (P = 0.245). PNN (WFA) was found to envelope GABAergic interneurons expressing the calcium-binding protein PV. We performed costaining for PV and specific proteins. PV expression was not detected in cortical neurons at P8 to P11 but was detected in the soma of scattered nonpyramidal neurons around P14. The mean density of PV-expressing cells increased substantially between P14 and P38 (Fig. 4C, P < 0.001, one-way ANOVA) and reached a stable level around P38 (Fig. 4C, P(P38-P42) = 0.647, P(P42-P60) = 0.085, one-way ANOVA). Correlation analysis revealed that the number of PNN-wrapped neurons and number of PV(+) interneurons were significantly negatively correlated with the number of microglia from P21 to P60 (Figs. 4D, 4E; R2 = 0.4062\0.3179, P < 0.0001; Pearson correlation coefficients). However, not all PV(+) interneurons were wrapped by PNNs. Only approximately 81% of PV(+) interneurons were wrapped by PNNs (Fig. 4F). 

Figure 4.

 

The number of PNNs and PV(+) interneurons increased rapidly beginning in the third week, and this increase was maintained from P35 to P42 to adulthood. (A) Example images of the V1B region stained for WFA (PNNs) and PV(+) interneurons in normal mice. (B, C) Changes in the density (mean ± SEM) of WFA-stained cells and PV(+) interneurons in V1B from P8-P60 (ns: P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA). (D–F) Correlation analysis of the numbers of PNNs and PV(+) interneurons with the number of microglia.

Briefly, these results indicate that the numbers of both PNNs and PV(+) inhibitory neurons follow similar patterns during development. The number of neurons surrounded by PNNs increases from postnatal day 14 (P14) to P42 and then remains constant. Similarly, the density of PV(+) interneurons tends to increase from P14 to P38 and then stabilizes. 

Microglia Elimination by a CSF1R Inhibitor Increases the Numbers of PNNs and PV(+) Interneurons in the Visual Cortex

Our results indicated that there are opposing trends between the number of microglia and the numbers of PNN-wrapped neurons and PV(+) interneurons from P21 to P42, with the critical period of mouse visual development occurring between P21 and P35. We hypothesized that microglia regulate the changes in the number of PV(+) interneurons during development. To verify our hypothesis, we treated mice with the CSF1R inhibitor PLX3397 for 2 weeks beginning at P14 to eliminate approximately 90% of microglia (Figs. 5A, 5B) 

Figure 5.

 

Changes in the density of PNNs and PV(+) interneurons after the depletion of microglia. (A) Timeline for assessing the extent of microglial depletion after PLX3397-containing chow administration. (B) Representative images of the V1B region stained for Iba1 following PLX3397-containing chow feeding. (C) Quantification of the microglial density during PLX3397 administration. PLX3397 exerted a dramatic effect on the number of microglia. Seven days of PLX3397 feeding led to a 42% reduction in microglial density, whereas an 88% reduction was found at the end of the 14-day PLX3397 treatment (P28). Microglia repopulated the visual cortex one week after PLX cessation. Two-way ANOVA with Sidak's multiple comparison test was used. (D, G) Example images of the V1B region stained for WFA (PNNs) and PV(+) interneurons following normal and PLX3397-containing chow feeding. (E) Comparison of the density of PNNs following normal and PLX3397-containing chow feeding. (F) Regression analysis of the numbers of PNNs and microglia following PLX3397-containing chow feeding. (H) Comparison of the density of PV(+) interneurons following normal and PLX3397-containing chow feeding. (I) Regression analysis of the numbers of PV(+) neurons and microglia following PLX3397-containing chow feeding. (J, K) Western blotting verified that microglial depletion and repopulation affected PV protein expression. (L) Representative images of the V1B region stained for VGLUT at P28 and P35 following normal and PLX3397-containing chow feeding. (M) Quantification of the number of VGLUT(+) neurons showed that PLX3397 did not affect the VGLUT(+) neuron density. *P < 0.05, **P < 0.01, ***P < 0.001

After one week of PLX3397 treatment (Supplementary Fig. S7A), at P21, the number of microglia was decreased by 40%. After two weeks of treatment, at P28, we observed a significant reduction in the microglial population in the mouse visual cortex of approximately 90% (P < 0.001, Figs. 5B, 5C). Astrocytes are also known to play a crucial role in the development of V1, including synaptic pruning. To confirm that PLX3397 specifically removes microglia and not astrocytes, we utilized an anti-GFAP antibody to mark astrocytes in the visual cortex at P28 and P35 (Supplementary Fig. S7B). We observed no change in the density of astrocytes after two weeks of PLX3397 treatment (P = 0.13) or one week after the discontinuation of PLX3397 treatment (P = 0.90, Supplementary Fig. S7C). Interestingly, at P35, one week after the discontinuation of PLX3397 treatment, the number of microglia was approximately double that in mice fed a regular diet (P < 0.001, Fig. 5C). This increase in the number of microglia may be attributed to increased microglial reactivity after stopping PLX3397 treatment. Even two to four weeks after discontinuation of PLX3397 treatment, the number of microglia remained twice that in mice fed a normal diet. 

The number of PNN-wrapped neurons in the visual cortex was increased by approximately 30% in mice treated with PLX3397 for one week (P21) compared to mice in the control group. Moreover, after 2 weeks of PLX3397 treatment (P28), the number of PNN-wrapped neurons in the visual cortex of the mice increased by approximately 75% (Figs. 5D, 5E) 

Similarly, the number of PV(+) interneurons in the visual cortex increased by approximately 20% in mice treated with PLX3397 for two weeks (P21) compared to mice in the control group (as shown in Figs. 5G, 5H). According to our regression analysis, there may be a causal relationship between microglial elimination and the recovery of PNNs and PV(+) interneurons (Figs. 5F, 5I). Increased PV protein expression at P28 and P35 was verified through western blotting (Fig. 5J). One week after stopping the PLX3397 treatment (P35), PV expression was elevated in the group previously treated with PLX3397 compared to the control group (Fig. 5K). At The same time, the proportions of PNN-positive PV neurons remained stable (Supplementary Fig. S7D). To confirm the effect of microglia on excitatory neurons, we labeled glutamatergic neurons with an anti-VGLUT1 antibody (Fig. 5L). Unlike the number of PV(+) inhibitory neurons, the glutamatergic neuron density was not changed by microglial elimination (Fig. 5M, P(P28) = 0.61, P(P35) = 0.82). The results showed that microglial elimination upregulates the expression of PV and PNN and that the repopulation of microglia reduces the numbers of PV(+) interneurons and PNNs. 

Elimination of Microglia Affects Ocular Dominance Plasticity During the Critical Period of Visual Development In Mice

Next, we investigated the impact of microglial elimination on visual cortex plasticity during the critical period. To accomplish this goal, we used the ocular dominance plasticity paradigm to test changes in developmental cortical plasticity. 

Mice in the experimental group were treated with PLX3397 from P14 to P35 were subjected to right-eye MD at P28 for seven days. Mice fed a normal diet were also subjected to MD at P28 for seven days. A homemade platinum-iridium wire electrode was implanted into the visual cortex contralateral to the deprived eye as the recording electrode, and another electrode was implanted into the contralateral frontal lobe as a reference electrode. VEPs were recorded at P35 (Fig. 6A). In normally reared mice, the contralateral eye-evoked response was two to three times greater than the nondeprived, ipsilateral, eye-evoked response, yielding a contralateral (C) to ipsilateral (I) eye ratio of ∼2.94. After seven days (P28 to P35) of MD during the critical period of visual development, the C/I ratio shifted significantly to favor the nondeprived ipsilateral eye, and the contralateral eye-evoked response was significantly decreased in normally reared mice, with a C/I ratio of ∼1.27 (Figs. 6B, 6C; P < 0.001). This was due to strong ocular dominance plasticity during the critical period of visual development in mice. However, both the VEP amplitude and the C/I ratio of the right eye of the PLX3397-treated and MD mice were greater than those of the right eye of the MD mice (P = 0.005) but did not reach the levels in the normal controls (Figs. 6B, 6C, P < 0.001). Treatment with PLX3397 partially prevented the shift in ocular dominance caused by MD for seven days during the critical period of visual development. Our results suggest that microglial depletion by PLX3397 treatment promotes a decrease in OD plasticity in animals during critical periods of visual development. 

Figure 6.

 

Depletion and repopulation of microglia alter ocular dominance plasticity and visual acuity in adolescent mice. (A, E) Timeline for assessing visual performance following PLX3397-containing chow administration. (B–D) At P35, compared with MDP28-35 group mice, PLX3397+MDP28-35 group mice showed a significant increase in the contralateral (closed-eye) VEP amplitude, visual acuity and C/I ratio following MD. (F–H) At P42, compared with MDP35-42 group mice, PLX3397STOP+MDP35-42 group mice showed a significant decrease in the contralateral (closed-eye) VEP amplitude, visual acuity and the C/I ratio following MD. *P < 0.05, **P < 0.01, ***P < 0.001; one-way ANOVA.

Then, we assessed the visual acuity of each group of mice. We determined visual evoked potential spatial frequency (SF) limits by linear regression and extrapolation of significant VEP sizes in the lower limbs at optimal SFs. By regression calculations, we found that seven days (P28 to P35) of MD during the critical period resulted in a 50% decrease in the SF limit in the deprived eye (Fig. 6D, P < 0.001), similar to deprivation amblyopia. However, seven days of MD led to a significant shift in the VEP SF limit in the deprived eye in the PLX33397-treated group compared with the normal diet group (Fig. 6D; P < 0.001). 

We aimed to confirm the reversibility of the effect of microglial elimination by PLX3397, and we observed a onefold increase in the number of microglia in the visual cortex through immunofluorescence staining one week after stopping PLX3397 treatment compared to that in normally reared mice. This raises the question of whether an increase in the number of microglia can restore ocular dominance plasticity after the critical phase. Mice were subjected to MD for seven days (P35 to P42) after stopping PLX3397 treatment (P35), and VEPs were recorded at P42 (Fig. 6E). In the group of mice that were previously treated with PLX3397, the ratio of responses for the nondeprived ipsilateral eye significantly increased after the treatment was stopped for two weeks (Fig. 6G, P = 0.0011), and the ratio of responses for the contralateral eye were significantly reduced compared to those of normally reared mice after seven days of MD (Fig. 6F; P < 0.001). These results suggested that microglia are important for ocular dominance plasticity. We also assessed the visual acuity of each group of mice. By regression calculations, we found that seven days (P35 to P42) of MD after the critical period resulted in a 23% reduction in the SF limit of the deprived eye (Fig. 6H; P = 0.005). However, in combination with PLX33397 treatment from P14 to P28, MD for seven days (P35 to P42) resulted in a significant reduction in the SF limit of the deprived eye compared with that in the normal diet group (Fig. 6H; P = 0.044). Overall, our data indicated that eliminating most microglia in the visual cortex clearly impairs ocular dominance plasticity in mice during the peak of the critical period. 

Microglia Ablation During the Critical Period Alters Spontaneous Synaptic Activity in the Visual Cortex

To determine the mechanism underlying these phenomena, we investigated how microglial depletion affects neuronal function in the visual cortex. We conducted whole-cell patch clamp recordings in the binocular region of V1 (bV1) in vehicle (Ctrl)- and PLX3397-treated mice at P28-30. The mEPSCs were recorded from L2/3 pyramidal neurons voltage clamped at −55 mV. 

We found that PLX3397 significantly reduced the mEPSC amplitude (Fig. 7E, Ctrl, 9.12 ± 0.20 pA; N = 4800 events/16 cells/6 mice; PLX3397, 8.38 ± 0.18 pA; N = 5100 events/17 cells/6 mice; t test, P = 0.01; K-S test, P > 0.05), and no change in mEPSC frequency was detected (Fig. 7C, Ctrl, 11.77 ± 0.86 Hz, N = 16 cells/5 mice; PLX3397, 11.86 ± 1.03 Hz, N = 17 cells/6 mice; P = 0.95; K-S test, P > 0.05). Next, we recorded mIPSCs in L2/3 bV1 pyramidal neurons. Contrary to previous results, a significant increase in the mIPSC frequency (Fig. 7D, Ctrl, 8.22 ± 0.60 Hz, N = 5400 events/18 cells/6 mice; PLX, 11.47 ± 0.64 Hz, N = 6000 events/20 cells/6 mice; p < 0.001; K-S test, p > 0.05) and a reduction in mIPSC amplitude (Fig. 7F, Ctrl, 27.33 ± 1.22 pA, N = 18 cells/6 mice; PLX, 19.37 ± 0.96 pA, N = 20 cells/6 mice, P < 0.001; K-S test, P > 0.05) was observed. 

Figure 7.

 

Microglia elimination from P28-35 altered neuronal function in the visual cortex. (A, G) Examples of mEPSCs recorded from L2/3 pyramidal neurons voltage clamped at −55 mV. (B, H) Examples of mIPSCs recorded from L2/3 pyramidal neurons. (C, D) PLX3397 did not alter the mEPSC frequency but increased the mIPSC frequency. (E, F) PLX3397 reduced the mEPSC amplitude and mIPSC amplitude. (I–L) No changes were found in the amplitude or frequency of mEPSCs or mIPSCs in the contralateral or ipsilateral V1B region after MD or PLX3397 treatment during the critical period. *P < 0.05, ***P < 0.001, ****P < 0.0001; t test.

To exclude the possibility that the changes in the frequency and amplitude of mEPSCs and mIPSCs were caused by MD, we recorded mEPSCs and mIPSCs in the contralateral and ipsilateral brains of mice in the PLX3397 group after MD (PLX P35), and the results showed that there was no significant change in the frequency of mEPSCs in the contralateral and ipsilateral brains (Fig. 7I, contralateral, 9.33 ± 0.90 Hz; N = 2400 events/8 cells/6 mice; ipsilateral, 9.05 ± 0.32 Hz; N = 3300 events/11 cells/6 mice; t test, P = 0.82; K-S test, P > 0.05), no change in the mEPSC amplitude (Fig. 7K; contralateral, 8.32 ± 0.19 pA, N = 7 cells/6 mice; ipsilateral, 19.37 ± 0.96 pA, N = 20 cells/6 mice, P = 0.09; K-S test, P > 0.05), no change in the mIPSC frequency (Fig. 7J; contralateral, 11.05 ± 1.62 Hz, N = 2100 events/7 cells/6 mice; ipsilateral, 9.19 ± 1.70 Hz, N = 2700 events/9 cells/6 mice, t test, p = 0.45; K-S test, P > 0.05; K-S test, P > 0.05), and no change in the mIPSC amplitude (Fig. 7K; contralateral, 22.39 ± 2.60 pA, N = 8 cells/6 mice; ipsilateral, 21.84 ± 1.29 pA, N = 11 cells/6 mice, P = 0.84; K-S test, P > 0.05). 

Thus the reduction in mEPSC amplitude and increase in mEPSC frequency may be explained by an increased number of immature synapses with lower AMPA receptor levels. Additionally, the decrease in mIPSC amplitude and increase in mIPSC frequency may indicate an increased number of presynaptic PV(+) interneurons. Taken together, these data suggest that depletion of microglia during the critical period affects synapse maturation and potentially alters visual cortex circuit connectivity. 

Discussion

Our study reveals that during the developmental process in the V1 region, the number of microglia increases in the second postnatal week, consistent with previous reports,^19,47 and reaches a peak around the third postnatal week. Subsequently, microglial numbers begin to decline at postnatal day 21 and stabilize throughout adulthood until six weeks of age. The morphology of microglia gradually transitions from an activated state to a resting state between P11 and P35. GO analysis based on snRNA-seq data indicated that, at eye opening (P14) and the peak of the critical period (P28), the pathways enriched in microglia were predominantly related to the promotion of synapse organization and glial activation. The immunofluorescence staining observed decrease in microglial numbers may be associated with the developmental increase in the expression of parvalbumin and PNNs instead of new neuron formation, as previously reported.^48–50 Additionally, eliminating microglia during the critical period of visual development with PLX3397 led to partial defects in ocular dominance, an effect that was reversed after discontinuing PLX3397 treatment. Moreover, the reduction in mEPSC amplitude, decrease in mIPSC amplitude, and increased frequency in V1 were closely correlated with a decline in experience-dependent plasticity, as indicated by previous research.⁵¹ Therefore the regulation of parvalbumin and PNNs by microglia is crucial for the development, connectivity, and plasticity of synapses in the visual cortex during mouse cortical development. 

In V1, activity-dependent development during critical periods is driven by the maturation of local inhibitory connections and late-developing interneurons.⁵² We observed that the decline in microglial numbers coincided with the end of the critical period and CNS maturation, with a strong correlation between microglial numbers, PV(+) interneurons, and PNNs.^22,53,54 Guided by snRNA-seq data, we explored the potential involvement of microglia in axonal pruning and synaptic plasticity, with a particular focus on their interactions with PV(+) interneurons. We found that microglial depletion increased PNN and PV(+) interneuron densities in the adolescent visual cortex and disrupted ocular dominance plasticity (ODP) during the peak critical period, as evidenced by changes in excitatory and inhibitory synaptic functions in V1B neurons during abnormal visual experiences.^{17,18,48,49,55–57} These findings underscore the role of microglia as homeostatic regulators of PNNs in the adult brain.⁵⁰ 

The CSF1R inhibitor PLX3397 effectively eliminated microglia without affecting astrocyte numbers, another glial cell type involved in synaptic pruning.⁵⁷ Microglial depletion led to an increase in perisynaptic ECM density and the number of perisomatic, predominantly GABAergic, synapses enveloped by PNNs. Our findings demonstrate that chronic microglial depletion over two weeks increases PNN density in the adolescent visual cortex, supporting the role of microglia in ECM remodeling.^48–50 PNNs, particularly those surrounding PV(+) interneurons, are typically associated with reduced plasticity and restricted synapse growth.^58,59 The observed increase in PNNs may be a compensatory response to heightened synaptic activity, as both in vitro and in vivo studies have shown that PNNs form in response to neuronal activity, with increased PV expression linked to enhanced neuronal activity.^59,60 

Contrary to our findings, another study reported no overall increase in PNN numbers after microglial elimination but did observe significant alterations in PNN organization, possibly because of sample size and the specific hippocampal region studied.⁶¹ Furthermore, inhibiting microglial phagocytosis or depleting microglia pharmacologically reduces the clearance of ECM components, such as aggrecan, leading to increased ECM deposition and PNN accumulation at synapses.^48,49 The maturation and organization of PV(+) interneurons are crucial for regulating the progression of the critical period.^52,62,63 Our results suggest that microglia regulate the maturation of PV(+) interneurons and critical period progression by controlling PNN deposition. Notably, microglial depletion primarily affects inhibitory neurons rather than glutamatergic excitatory neurons, consistent with previous findings showing no change in VGLUT1 puncta density after microglial elimination and repopulation.⁶¹ 

Microglial depletion and repopulation modulate neural activity in the developing visual cortex. Using the CSF1R inhibitor PLX3397, we explored the role of microglia in ODP. Our findings support previous studies indicating that microglia are essential for ODP. Depletion primarily affects inhibitory neurons, disrupting the excitatory/inhibitory balance and impairing ODP during the critical period, as shown by MD-induced shifts in ocular dominance using VEPs. This aligns with reports that deletion of the P2Y12 gene blocks ODP and that PLX3397 treatment inhibits ODP.⁶⁴ Although PLX3397 also affects OPCs,⁶⁵ its broader biological effects may uncover complex interactions between microglia and other cells, which could be missed with single-target inhibitors. A rebound in microglia numbers after PLX3397 removal is consistent with previous findings reported by multiple groups, highlighting that this phenomenon is not unexpected.^66,67 This suggests that repopulated microglia may play a role similar to resident microglia, although their gene expression profiles and functional properties may exhibit certain distinctions. Microglial phagocytic activity, modulated by visual experience,⁶⁸ and their interactions with PV(+) interneurons are crucial for maintaining excitatory/inhibitory balance during the critical period.^18,69–72 The extracellular matrix (ECM) also plays a key role in synaptic stabilization, with ECM proteins interacting with transmembrane proteins to influence synaptic structure and vesicle release.^73–77 The microglia repopulation plays a critical role in maintaining neural homeostasis and facilitating repair, functioning similarly to resident microglia and inhibiting extracellular matrix remodeling.^78–80 Additionally, repopulated microglia promote brain repair and neuroprotection, aiding recovery from neural injuries and cognitive deficits.⁸¹ Consistent with previous studies, we found that microglial depletion delayed the maturation of PV(+) interneurons and partially impaired ocular dominance plasticity during the critical period of visual development, while microglial repopulation may enhance synaptic plasticity.⁵ Although our study highlights the potential role of microglial repopulation in restoring synaptic function, the gene expression profile of repopulated microglia has not yet been thoroughly analyzed using snRNA-seq. Future research should employ snRNA-seq to further investigate which functional subpopulations these repopulated microglia belong to, as well as to elucidate their signaling interactions with PV(+) interneurons and excitatory neurons. This will provide valuable insights into how repopulated microglia contribute to neural circuit remodeling, particularly through their interactions with PV neurons and excitatory neurons. 

Electrophysiological recordings from this study underscore the role of microglia in homeostatic plasticity. The mEPSC and mIPSC recorded in the V1 region of the control group were in agreement with findings from previous studies.⁸² Microglial depletion increases PV expression and the number of PV(+) interneurons, reducing cortical excitability. This depletion also disrupts synaptic organization and signaling to PV(+) interneurons, as evidenced by decreased mEPSC and mIPSC amplitudes, indicating reduced GABAergic receptor expression. The resulting increased inhibition and decreased cortical activity leads to impaired ocular dominance plasticity. Whole-cell patch clamp recordings further confirm that microglial depletion during the critical period disrupts excitatory and inhibitory synaptic functions, likely through mechanisms involving spine pruning and remodeling of PV(+) interneurons and PNNs. 

However, the role of microglia in ODP remains controversial, as other studies have reported conflicting results. For example, Schecter et al.⁸³ observed that the deletion of the microglial gene CX3CR1 does not block ODP. Likewise, Lowery et al.⁸⁴ reported that CX3CR1 is not essential for ODP, and Welsh et al.⁸⁵ found that the key microglial signaling molecule C1q is not required for ODP. Moreover, our findings also differ from those reported by the McGee lab,⁸⁶ which used multiunit recordings and calcium imaging to study OD plasticity. These differences underscore the diversity of methodological approaches in the study of cortical plasticity, each offering distinct perspectives. Although multiunit recordings and calcium imaging provide a more direct measure of neuronal activity, VEPs offer a unique perspective by capturing broader cortical responses. It is also important to consider the potential impact of the drug used in this study. PLX3397, a CSF1R inhibitor, is known to have broader off-target effects compared to PLX5622, which exhibits greater specificity for CSF1R.^65,87–89 These off-target effects may influence other cell types, such as oligodendrocyte progenitor cells and peripheral immune cells, including bone marrow-derived cells. Such broader effects could indirectly contribute to the observed differences in ODP and complicate the interpretation of microglia-specific roles in cortical plasticity. This highlights the need for caution when interpreting results obtained with PLX3397, because its systemic biological effects might obscure microglial-specific contributions. In future studies, incorporating alternative approaches, such as using the more selective PLX5622 or employing genetic tools, could help disentangle microglial-specific functions from off-target effects and provide further clarity on their roles in cortical plasticity. Additionally, our snRNA-seq analysis predicted microglia-neuron interactions based on computational modeling; however, these predictions do not confirm the actual presence of such interactions in vivo. Despite this limitation, the predicted interactions serve as a valuable hypothesis-generating tool, offering insights into potential microglial influences on PV(+) interneurons and guiding future research directions. To validate these interactions, future studies could employ direct experimental approaches, such as two-photon in vivo imaging, to observe microglial contact and synaptic remodeling dynamics in real-time. This would provide a more definitive understanding of how microglia contribute to synaptic plasticity and ODP. Looking forward, we recognize that integrating additional techniques, such as multiunit recordings or calcium imaging, would further validate and extend our findings. This integrative approach has the potential not only to reconcile differences between studies but also to advance our understanding of how sensory deprivation and recovery shape cortical circuits. 

In conclusion, our study revealed that microglia play a crucial role in regulating neural circuit connectivity and activity by modifying both excitatory and inhibitory synaptic connections to excitatory neurons in the developing visual cortex, especially inhibitory synaptic connections. These findings open up new potential avenues for increasing neural plasticity in adulthood, which could be applied in the treatment of brain disorders and injuries, including neurodegenerative and neuropsychiatric disorders. 

Acknowledgments

Supported by the National Natural Science Foundation of China (No.81300791), the Tianjin Health Research Project(Nos. TJWJ2021MS042), Key Project of Tianjin Eye Hospital (No. YKZD2004), the third Tianjin Talent Development Program and the High-level Talents Program in TJHS Tianjin Key Medical Specialty Construction Project, The Open Fund of the Institute of Vision Science&optometry, Nankai University (NKSGY202301, NKSGZ202302), Tianjin Science and Technology Program Project (24JRRCRC00150) and Tianjin Key Medical Discipline Construction Project (No. TJYXZDXK-016A). 

Disclosure: X. Wang, None; K. Li, None; L. Guo, None; X. Liu, None; Y. Guo, None; W. Zhang, None