Early‐life IL‐4 administration induces long‐term changes in microglia in the cerebellum and prefrontal cortex

Abstract

Microglia are crucial for brain development and their function can be impacted by postnatal insults, such as early‐life allergies. These are characterized by an upregulation of interleukin (IL)‐4 levels. Allergies share a strong comorbidity with Autism Spectrum Disorders (ASD) and Attention‐Deficit/Hyperactivity Disorder (ADHD). We previously showed that early‐life allergic asthma induces hyperactive and impulsive behaviors in mice. This phenotype was reproduced in animals administered with IL‐4 in the second postnatal week. Mechanistically, elevated IL‐4 levels prevented microglia‐mediated engulfment of neurons in the cerebellum, resulting in a surplus of granule cells and consequent dysfunction in cerebellar connectivity. Here, we aimed to further understand the impact of early IL‐4 administration in microglia of the cerebellum and the prefrontal cortex (PFC), two brain regions with protracted developmental programs and susceptible to immune system malfunction after birth. While IL‐4 administration induced differential short‐term effects on microglia in the cerebellum and PFC, both regions presented similar microglial features in adult mice. Although Sholl analysis did not reveal significant alterations in overall microglia morphology at postnatal day (P)10, the density of microglia was decreased in the cerebellum at this age, especially in the granular layer (GL), but remained unaltered in the PFC. Interestingly, the presence of microglia with phagocytic cups, morphological features important for whole‐cell engulfment, was decreased in both regions. When assessing the long‐term consequences of IL‐4 administration, cerebellar and PFC microglia were hypo‐ramified and exhibited increased overall density. Importantly, microglia alterations were exclusive to the GL of the cerebellum and the infralimbic region of the PFC. Our results show that postnatal elevated levels of IL‐4 impair the percentage of microglia engaged in cell clearing in two brain regions with protracted developmental programs. Interestingly, IL‐4‐exposed microglia adapt a similar phenotype in the adult cerebellum and PFC. Our data suggest that this early‐life increase in IL‐4 levels is sufficient to elicit long‐lasting alterations in microglia, potentially increasing cell susceptibility to later insults.

Elevated postnatal levels of Interleukin (IL)‐4, such as those observed in allergies, induce short‐ and long‐term alterations in microglia in the cerebellum and prefrontal cortex, two brain regions with protracted maturation. While the cerebellum is particularly affected in early life, this insult induces similar alterations in both brain regions in adulthood. IL‐4 administration decreases the percentage of microglia engaged in cell clearing in the postnatal period while, in adult mice, microglia are hypo‐ramified and present increased density. These effects are restricted to the granular layer of the cerebellum and the infralimbic region of the prefrontal cortex. Considering the comorbidity between allergies and Autism Spectrum Disorders and Attention‐Deficit/Hyperactivity Disorder, this work highlights how early‐life environmental triggers may impair microglia physiology throughout life and potentially contribute to the etiology of these conditions.

Abbreviations

ADHD

Attention‐Deficit/Hyperactivity Disorder

ASD

Autism Spectrum Disorders

Cg1

Anterior cingulate cortex

GL

Granular layer

IBA‐1

Ionized calcium‐binding adapter molecule 1

IL

Infralimbic

IL‐4

Interleukin‐4

ML

Molecular layer

MO

Medial orbitofrontal cortex

PCL

Purkinje cell layer

PFC

Prefrontal cortex

PrL

Prelimbic

1. INTRODUCTION

Microglia are resident immune cells of the brain, essential in sculpting neuronal networks (Guedes et al., 2022). Among other functions, microglia regulate the number of neurons by performing neuronal engulfment and inducing apoptosis (Cunningham et al., 2013; Marín‐Teva et al., 2004), but also by aiding in the migration of neuronal precursors (Squarzoni et al., 2014). Classically, pathological microglial activation has been associated with a transition from a ramified to an amoeboid‐like phenotype. However, the morphological state of these cells is complex, and correlating morphology with function has proven to be a challenge. Adding to the physiological changes across development in these cells, insults such as stress (Delpech et al., 2016), infection (Mattei et al., 2017), or allergies (Saitoh et al., 2021) can impact microglia morphology and function, leading to a disruption of brain development.

As one of the main mediators of allergic responses, IL‐4 is essential for the development of atopic dermatitis, allergic asthma, and rhinitis. The role of IL‐4 in the peripheral immune system is well described (Keegan et al., 2021), but this cytokine is also present in the brain milieu (Guedes et al., 2023; Wang et al., 2024). Here, it not only modulates microglia (Fenn et al., 2014; Francos‐Quijorna et al., 2016), but also plays a role in memory and learning (Derecki et al., 2010; Herz et al., 2021). However, most studies on IL‐4 on microglia have been conducted in models of traumatic injury (Liu et al., 2016; Ting et al., 2020; Zhao et al., 2015), with few exploring the effects of this cytokine during neurodevelopment (de Araujo‐Martins et al., 2013; Wang et al., 2024).

Autism Spectrum Disorders (ASD) and Attention‐Deficit/Hyperactivity Disorder (ADHD) are the most common neurodevelopmental conditions and can be predisposed by several environmental stressors, especially if exposure occurs during the embryonic and postnatal developmental periods (Estes & McAllister, 2016; Faraone et al., 2021). Epidemiological studies suggest that there is a strong comorbidity between allergies and ASD (Xu et al., 2018), and ADHD (Cortese et al., 2018), and the severity of allergic and ADHD symptoms is positively correlated (Chuang et al., 2022). Treatment with dupilumab (an antibody against IL‐4 receptor) for atopic dermatitis decreased the need of ADHD medication in children with both conditions (Yildirim et al., 2023), while allergy medication has also proven efficient in ameliorating ADHD behaviors in patients (Ding & Lu, 2024; Suntiwes et al., 2023; Thamrongsak et al., 2022).

Previously we showed that, similarly to the induction of early‐life allergic asthma, peripheral injection of recombinant IL‐4 in the postnatal period reduces elimination of cerebellar granule cells through a mechanism of microglia‐mediated neuronal engulfment (Guedes et al., 2023). Consequently, there is a surplus of granule cells in IL‐4‐injected mice together with deficits in neuronal connectivity and hyperactive–impulsive behaviors (Guedes et al., 2023). Despite lack of conclusive data, cerebellar and prefrontal cortex (PFC) dysfunction have been associated with both ASD (D'Mello & Stoodley, 2015; Leisman et al., 2023) and ADHD (Curatolo et al., 2010; Kasparek et al., 2015). These two brain regions share a protracted developmental program that extends well into the postnatal period (Chini & Hanganu‐Opatz, 2021; Sathyanesan et al., 2019), making them susceptible to immune system challenges after birth. In this work we probed how high levels of IL‐4 during the second postnatal week influence microglia in the cerebellum and the PFC. We observed that, in the short term, this cytokine reduced the percentage of microglia presenting phagocytic cups, a morphological structure essential for whole‐cell engulfment, without inducing overt changes in microglia ramification. However, in adult mice, microglia of IL‐4‐injected animals displayed long‐term morphological alterations characterized by hypo‐ramification, increased IBA‐1 expression, and higher density in a layer‐specific manner. Our results show that an IL‐4 insult in early ages produces acute and chronic alterations in microglia, which may influence their susceptibility to subsequent environmental insults and compromise their homeostatic function.

2. MATERIALS AND METHODS

2.1. Mouse lines and treatments

Experimental animals—PV::tdT mice—heterozygous for both alleles were generated by breeding Parvalbumin‐Cre mice (knock‐in line from Silvia Arber) and Rosa26 loxP‐Stop‐loxP tdTomato mice (Ai9). This model presents parvalbumin‐positive neurons labeled with tdTomato. Mouse cages were maintained in a controlled environment at 22°C and 60% humidity under a 12 h light/dark cycle in a ventilated rack system. Pups were kept with the mother until weaning (postnatal day (P)21), after which they were separated by sex (4 animals per cage). Animals were given food and water ad libitum. Mice were either administered with saline solution (0.9% NaCl) or IL‐4 (0.1 mg/kg, PeproTech, Cat# 214‐14; Accession number: P07750) intraperitoneally at P8 and sacrificed at P10 for evaluation of short‐term effects, or at P8 and P13 and sacrificed at P60 for long‐term assessments. IL‐4 peripheral administration in mice was previously shown to be sufficient to induce a complex inflammatory reaction resembling that observed in human allergic disease (Tepper et al., 1990) and characterized by bronchoalveolar lavage fluid eosinophilia, airway hyperresponsiveness, and goblet cell hyperplasia (Perkins et al., 2006). Timepoints for analysis were chosen based on the importance of the first two postnatal weeks for the formation and maturation of the cerebellum (Leto et al., 2016) and prefrontal cortex (Chini & Hanganu‐Opatz, 2021). Animals from both sexes were used for all experiments and assigned to experimental groups arbitrarily, i.e., male and female littermates were injected with saline or IL‐4 arbitrarily (sex‐balanced experimental groups). Although no sample size calculation was performed, the number of animals used in each experimental group was based on previous experience (Guedes et al., 2023). Experimenters were blinded to the experimental condition before conducting any image acquisition or analysis. Maintenance and handling were performed in compliance with all ethical regulations for animal testing and research, including FELASA Animals Use and Care Guidelines and European Directives on Animal Welfare. All experiments were conducted in accordance with protocols approved by ORBEA (Institutional Animal Welfare Body of the University of Coimbra/CNC, reference number 184/2018 and ORBEA_282_2021/10032021) and DGAV (Portuguese Regulatory Agency).

2.2. Immunohistochemistry

Animals were deeply anesthetized by being placed in a chamber with 5% isoflurane (Abbot) dispersed in air, followed by perfusion with Phosphate Buffered Saline (PBS). The brain was quickly removed and post‐fixed in 4% paraformaldehyde overnight, followed by incubation in a 30% sucrose in PBS solution and storage at −80°C until processing. A cryostat (CryoStar NX50, Thermo Fisher Scientific, USA, RRID:SCR_022732) was used to serially cut 50 μm sagittal (cerebellum) or coronal (PFC) sections, which were collected to multiwell plates with 0.02% azide in PBS and stored at 4°C until further use. For immunohistochemistry analysis, 6 slices per animal equally spaced by 300 μm were selected from the cerebellar vermis or PFC region and processed in 12 multiwell plates (3 sections/well). Sections were washed thrice with PBS for 10 min to remove azide, followed by permeabilization and blocking in a solution of 10% normal goat serum (Sigma‐Aldrich, Cat# G9023) and 0.1% Triton (Thermo Fisher Scientific, Cat# 10671652) in PBS for 1 h at room temperature (RT). Slices were then incubated with primary Rabbit Anti‐IBA‐1 antibody (1:1000; Fujifilm Wako, Cat# 019‐19741; RRID:AB_839504) in the same blocking solution overnight at 4°C. On the next day, sections were washed three times with PBS for 15 min, followed by incubation with secondary antibody Alexa Fluor 488 Donkey Anti‐Rabbit IgG (1:250; Thermo Fisher Scientific, Cat# A32790; RRID:AB_2762833) for 2 h at RT in the dark. Sections were washed three more times with PBS for 15 min and mounted into gelatinized microscope slides. For slices used for 3D microglia reconstructions VECTASHIELD HardSet Mounting Medium with DAPI (Vector Laboratories, USA, RRID:AB_2336788) was used. For phagocytic cup representative images where nuclear fragments are visible, as well as for microglia density estimation, slices were first incubated in the dark for 5 min in 1 mg/mL of Hoechst (Thermo Fisher Scientific, Cat#H3570) in PBS, washed twice with PBS for 10 min), and mounted using Mowiol (Sigma‐Aldrich, Cat#81381) mounting medium. For the representative images of microglia stained with CD68, a primary Rat Anti‐CD68 antibody (1:500; Bio‐Rad, Cat# MCA1957; RRID:AB_322219) was used overnight followed by incubation with a secondary antibody Alexa Fluor 647 Goat Anti‐Rabbit IgG (1:250; Thermo Fisher Scientific, Cat# A21247; RRID:AB_141778) for 2 h at RT in the dark. All wash and incubation steps were performed under constant agitation.

2.3. Imaging and analysis

2.3.1. Confocal microscopy

For surface and filament analysis, IBA‐1‐stained mounted slides were imaged in a 710 LSM confocal microscope (Carl Zeiss, Germany, RRID:SCR_018063) using a PA 40× NA 1.4 DIC lens objective and Zen Black software (v2.6, Carl Zeiss, RRID:SCR_018163). For each animal, microglia from three brain sections were imaged, either in the cerebellum (lobules IV/V) or PFC. After locating the desired cell, a 1024 × 1024 px Z‐stack of 20–40 planes was acquired with a 0.6 μm step. Confocal microscope acquisition settings were the same across all conditions.

2.3.2. Microglia morphology analysis

Microglia were 3D reconstructed in Bitplane Imaris (v9.5, Oxford Instruments, UK, RRID:SCR_007370) using a 0.15 μm surface grain size (3 Saline mice and 3 IL‐4 mice in the cerebellum P10; 4 Saline mice and 4 IL‐4 mice in the PFC P10; 3 Saline mice and 3 IL‐4 mice in the cerebellum P60; 4 Saline mice and 4 IL‐4 mice in the PFC P60). Filaments were created on the surface‐masked channel using Imaris Autopath followed by manual inspection and correction. The surface area/volume ratio (SA:V), IBA‐1 staining intensity (obtained by normalizing the sum of the intensity of all points in the surface‐masked channel to the volume of the surface and presented as IBA‐1 intensity sum/volume (AU/μm³)) and sphericity parameters were obtained from the created surface, while Sholl analysis (number of Sholl intersections), max cell radius, and number of ramifications (sum of Sholl intersections) were extracted from the generated filaments. Convex hull (Territory), determined as the volume of the smallest 3D polygon that encapsulates the cell, was obtained using the 3D convex hull plugin (Sheets et al., 2011) in ImageJ Fiji (RRID:SCR_002285, RRID:SCR_003070 (Rueden et al., 2017; Schindelin et al., 2012)). This parameter was used as a correlate of the territory occupied by each microglia. Solidity was defined as the percentage of the convex hull occupied by a cell and was obtained by dividing the convex hull by the volume of the cell.

2.3.3. Microglia density estimation

For analysis of microglia density, slices were imaged in a Axio Imager 2 microscope (Carl Zeiss, Germany, RRID:SCR_018876) using a PA 20× NA 0.8 lens objective and the software Stereo Investigator (v2018.1.1, MBF Bioscience, USA, RRID:SCR_024705) (5 Saline mice and 5 IL‐4 mice in the cerebellum P10; 6 Saline mice and 6 IL‐4 mice in the PFC P10; 3 Saline mice and 3 IL‐4 mice in the cerebellum P60; 5 Saline mice and 5 IL‐4 mice in the PFC P60). For each slice, regions of interest were delineated. The cerebellum was divided into molecular layer (ML), Purkinje cell layer (PCL), and granular layer (GL), and the PFC was divided into anterior cingulate cortex (Cg1), prelimbic (PrL), infralimbic (IL) and medial orbitofrontal cortex (MO) and further subdivided into Layers II/III and V/VI. Microglia were then counted using the optical fractionator method, followed by determination of microglia density (dividing the number of counted microglia markers—cell body—by the respective assessed area). For the cerebellum, a sampling grid of 50 × 50 μm squares and a counting frame of 25 × 25 μm were used, while for the PFC a sampling grid of 150 × 150 μm and a counting frame of 100 × 100 μm were used. Microscope settings were maintained across all slices of the same age.

2.3.4. Determination of the percentage of microglia with phagocytic cup morphological structures

For the cerebellum, microglia images acquired for 3D reconstruction were probed for the presence of phagocytic cups (3 Saline mice and 3 IL‐4 mice). In the PFC, cells marked during the optical fractionator method were also assessed for the presence of phagocytic cups (6 Saline mice and 6 IL‐4 mice). Percentages were calculated with respect to the total number of assessed microglia for each region. Representative images of phagocytic cups were acquired in the confocal microscope using a PA 63× NA 1.4 DIC lens objective by capturing a Z‐stack with the same settings as before at 2048 × 2048 px and then reconstructed in Imaris for channel masking using the created surface.

2.4. Statistical analysis

Data are presented as mean values ± standard error of the mean (SEM) with individual points. Sholl analysis statistics were performed by employing a Quasi‐Poisson generalized additive mixed model (QS‐GAMM) to account for the nonlinear nature of the microglia's ramification profile. A quasi‐Poisson distribution was assumed as the response variable is inherently count‐based and non‐negative, also allowing for flexible estimation of the variance. The model accounted for the nested nature of the data in random effects, to control for pseudo‐replication issues, with measures at multiple radii from multiple cells within each animal. An autoregressive covariance structure (AR1) was added to the model to account for the correlation of the values measured at nearby radii. To model the nonlinear relationship between radius and the count of intersections, we employed thin plate splines. This approach allowed us to flexibly capture the underlying trend in the data that linear terms alone could not adequately represent. Goodness of fit was evaluated after exploring the models' residuals, Akaike Information Criterion and Bayesian Information Criterion measures. A linear mixed model (LMM) was fitted individually for the value of intersections at every radius, accounting for the nesting of different cells within each animal and obtained p‐values were corrected for multiple comparisons using the Sidak method. For radar plots and density analysis, a LMM was also fit for each parameter/layer, while accounting for the nested structure of different cells or slices from the same animal by including the animal ID as a random effect, and p‐values were corrected for multiple comparisons using the Sidak method. Data used in GAMMs and LMMs were log‐transformed to ensure a better model fit. For data on the percentage of microglia with phagocytic cups, a Binomial generalized LMM was fit due to the binary nature of the measure (presence/absence of a cup structure), accounting for the different animals per condition, using the animal ID as a random effect and corrected for multiple comparisons using the Sidak method. Statistical analysis was conducted in the R software (v4.4.1 (R Core Team, 2020), RRID:SCR_001905) using in‐house scripts. Code used for statistical analysis can be found at https://github.com/ncblcnc/IL‐4_project. Graphs were created using Graphpad Prism (v9.3, GraphPad Software, USA, RRID:SCR_002798). Outliers were analyzed using the ROUT method with Q < 0.5%, with one slice from one animal being excluded from Saline P60 microglia density assessment. p value reporting is accompanied by the estimate ($\widehat{\beta }$), standard error (SE), and t value for each parameter. Statistical significance was set as ***p < 0.001, **p < 0.01 and *p < 0.05.

3. RESULTS

3.1. Assessment of short‐term effects of IL‐4 on cerebellar and PFC microglia morphology

IL‐4 is a critical cytokine for the development of allergic airway inflammation (Tepper et al., 1990). To evaluate the impact of postnatal administration of IL‐4 in microglia in the cerebellum and PFC, two brain regions that present protracted circuit maturation (Chini & Hanganu‐Opatz, 2021; Leto et al., 2016), we divided the cerebellar cortex into the ML, PCL and GL (Haines & Dietrichs, 2012); the PFC was divided into the Cg1, PrL, IL and MO and further subdivided in layers II/III (more external, input layers) and V/VI (more internal, output layers) (Anastasiades & Carter, 2021). These subdivisions allowed for testing the hypothesis that exposure to IL‐4 could induce layer‐specific effects in both the cerebellum and PFC, which can potentially impact the function of different neurons. Both males and females were analyzed but, since no significant differences were observed between sexes, the data were not segregated. IL‐4‐induced short‐term effects on microglia were assessed in P10 mice, administered with IL‐4 at P8 (Figures 1 and 2). The main goal was to understand whether the effect of IL‐4 on the previously described microglia‐mediated engulfment of neurons (Guedes et al., 2023) was correlated with morphological alterations. In the cerebellum, IL‐4 administration induced a significant increase in microglia IBA‐1 intensity (Figure 1c, LMM with Sidak correction, $\widehat{\beta }$ = 0.6481, SE = 0.1178, t = 5.50, p < 0.001), while no significant alterations in other surface analysis parameters were observed. Sholl profile indicated a tendency for microglia to become hyper‐ramified (Figure 1d,e, Quasi‐Poisson generalized additive mixed model (QS‐GAMM), $\widehat{\beta }$ = 0.045, SE = 0.047, t = 0.944, p = 0.345), which achieved statistical significance in the ML (Figure 1f,i, QS‐GAMM, $\widehat{\beta }$ = 0.228, SE = 0.086, t = 2.664, p = 0.007). Furthermore, ML microglia of IL‐4‐injected animals acquired a morphological profile similar to those of untreated P15 animals (Figure S1), suggesting that microglia of this layer may be experiencing accelerated maturation in what concerns morphological features. No significant differences were observed in microglia from either the GL or PCL between saline‐ and IL‐4‐injected animals (Figure 1g–i).

FIGURE 1.

Effects of postnatal IL‐4 administration on cerebellar microglia morphology at P10. (a, b) Scheme of IL‐4 administration (a) and representation of the region of the cerebellum where microglia images were collected, as well as the different layers of the cerebellar cortex (b). (c) Surface analysis indicates that alhtough no significant alterations were found in microglia shape, complexity, or occupied territory, there is an increase in IBA‐1 expression in IL‐4‐injected animals. Saline/IL‐4 n = 33/40 cells from 3 mice per condition. p‐values from a linear mixed model per parameter with Sidak correction. Territory represents the cell 3D convex hull (μm³); ramifications represent the total number of Sholl intersections; IBA‐1 is calculated as IBA‐1 intensity sum/volume (AU/μm³); SA:V is the ratio between surface area and volume of the cell. (d) Animals administered with IL‐4 display a non‐statistically significant increase in Sholl intersections close to their soma. Sholl‐filled area represents a moving average smoothing of 5 μm intervals. Scale bar for representative traces 15 μm. Saline/IL‐4 n = 33/40 cells from 3 mice per condition. (e) Representative 3D reconstructions of microglia in both conditions (IBA‐1, microglia, green). Scale bar 5 μm. (f–h) When compartmentalizing Sholl analysis by layer, despite no significant alterations being observed in the PCL (g) and GL (h), microglia of the ML show a significant increase in the number of Sholl intersections in IL‐4‐injected animals (f). Sholl‐filled area represents a moving average smoothing at 5 μm intervals. Scale bar for representative traces 15 μm. GL Saline/IL‐4 n = 15/22, PCL Saline/IL‐4 n = 11/8, ML Saline/IL‐4 n = 7/10 cells from 3 mice per condition. p‐values from a Quasi‐Poisson generalized additive mixed model. (i) Representative images of microglia distribution in the different layers of the cerebellar cortex (IBA‐1, microglia, green). Scale bar 20 μm. Radar plot data are presented as mean ± SEM. *p < 0.05, ***p < 0.001 with respect to saline‐injected mice. GL, granular layer; ML, molecular layer; PCL, Purkinje cell layer.

FIGURE 2.

Microglia of IL‐4‐injected animals show no morphological alterations in the prefrontal cortex (PFC) at P10. (a, b) Scheme of IL‐4 administration (a) and representation of the PFC and different subdivisions where analysis was performed (b). (c, d) No significant alterations were found in microglia of the PFC following IL‐4 injection in either surface analysis (c) nor total Sholl analysis (d). (e) Representative 3D reconstructions of microglia in both conditions (IBA‐1, microglia, green). Scale bar 5 μm. Sholl‐filled area represents a moving average smoothing at 5 μm intervals. Scale bar for representative traces 15 μm. Saline/IL‐4 n = 32/32 cells from 4 animals per condition. Territory represents the cell 3D convex hull (μm³); ramifications represent the total number of Sholl intersections; IBA‐1 is calculated as IBA‐1 intensity sum/volume (AU/μm³); SA:V is the ratio between surface area and volume of the cell. (f‐h) Microglia from IL‐4‐injected animals display no alterations in the number of Sholl intersections in either external (f) or internal layers (g). Sholl‐filled area represents a moving average smoothing at 5 μm intervals. Scale bar for representative traces 15 μm. Saline/IL‐4 n = 16/16 cells per layer from 4 animals per condition. (h) Representative images of microglia in the different layers of the PFC in both conditions (IBA‐1, microglia, green). Scale bar 20 μm. Radar plot data are presented as mean ± SEM. Cg1, anterior cingulate Cortex; IL, infralimbic; MO, medial orbitofrontal cortex; PrL, prelimbic.

In the PFC, in contrast, IL‐4 administration did not induce overall changes in microglia morphology, either in surface analysis parameters (Figure 2c) or Sholl profile (Figure 2d,e), even after layer subdivision (Figure 2f–h). Of note, at P10, microglia of the PFC present striking intrinsic differences from those of the cerebellum (Figure S2a,c, QS‐GAMM, $\widehat{\beta }$ = 0.498, SE = 0.059, t = 8.493, p < 0.001), being characterized by a higher number of ramifications, thinner filaments (significantly decreased SA/V ratio), and an elongated profile (decreased sphericity) (Figure S2b, LMM for each parameter with Sidak correction; Ramifications—$\widehat{\beta }$ = 0.744, SE = 0.083, t = 8.899, p < 0.001; SA/V—$\widehat{\beta }$ = 0.457, SE = 0.077, t = 5.872, p < 0.001; Sphericity—$\widehat{\beta }$ = −0.029, SE = 0.007, t = −3.917, p = 0.002). Despite occupying a similar overall territory, because of the presence of thicker branches, cerebellar microglia showed a higher solidity (volume of the cell divided by its territory; Figure S2b, LMM for each parameter with Sidak correction, $\widehat{\beta }$ = −0.039, SE = 0.009, t = −4.399, p < 0.001).

3.2. Evaluation of microglia density in the cerebellum and PFC following IL‐4 exposure

To further investigate the effects of IL‐4 administration on the phenotype of microglia at P10, we probed both the cerebellum and PFC for alterations in cell density and in the percentage of microglia displaying phagocytic cups as a correlate of phagocytic activity (Figures 3 and 4). At P10, cerebellar microglia are not equally distributed between layers, being more concentrated in the GL (Figure 3c). We observed a significant decrease in total microglia density in IL‐4‐injected mice (Figure 3d inset, LMM, $\widehat{\beta }$ = −0.288, SE = 0.109, t = −2.623, p = 0.030) that was also statistically significant in the GL and, to a lesser extent, in the ML (Figure 3d, LMM for each layer with Sidak correction, GL—$\widehat{\beta }$ = −0.335, SE = 0.108, t = −3.092, p = 0.042; ML—E = −0.344, SE = 0.216, t = −1.593, p = 0.387). These results suggest that, although no differences in microglia morphology were observed in the GL (Figure 1h), the decrease in microglia density in this region may correlate with the increased density of granule cells previously observed at P10 in this model (Guedes et al., 2023). When assessing the percentage of microglia with phagocytic cups (Figure 3e,f), which label the lysosomal marker CD68 (Figure S3), we observed an overall decrease, although not statistically significant, in IL‐4‐injected animals (Figure 3f, Binomial generalized LMM (B‐GLMM), $\widehat{\beta }$ = −0.471, SE = 0.449, z = −1.049, p = 0.501). Of note, most microglia displaying phagocytic cups are present in the GL in both conditions (Figure 3f), which emphasizes the role of microglia in the elimination of granule cells at P10.

FIGURE 3.

IL‐4‐injection alters microglia density in the cerebellum at P10. (a) Scheme of IL‐4 administration. (b) Low magnification representative image of the lobule of the cerebellar cortex from where microglia were analyzed (IBA‐1, microglia, green), subdivided by layers. Scale bar 50 μm. (c) High magnification representative image of the distribution of microglia in the different cerebellar cortex layers. Scale bar 10 μm. (d) IL‐4 injection at P8 induces a significant reduction of microglia density in the cerebellar cortex at P10, which is pronounced in the GL. Saline/IL‐4 n = 13/14 slices from 5 animals per condition (males are represented as squares and females as diamonds). p‐values from a linear mixed model for total density (inset) and per layer with Sidak correction for multiple comparisons. (e) Representative images of a microglia with a phagocytic cup staining Hoescht, as well as a 3D reconstruction and cup close‐up (IBA‐1, microglia, green; Hoechst, DNA, blue). (f) Overall, there is a tendency for a reduction of the percentage of microglia with phagocytic cups after IL‐4 administration. Saline/IL‐4 n = 33/40 cells from 3 mice per condition. Density data are presented as mean ± SEM. *p < 0.05 with respect to saline‐injected mice. GL, granular layer; ML, molecular layer; PCL, Purkinje cell layer.

FIGURE 4.

IL‐4 induces a significant decrease in the percentage of microglia with phagocytic cups at P10 in the PFC. (a, b) IL‐4 administration protocol (a) and low magnification representative image of the layer subdivision of the PFC (b). Scale bar 20 μm for the close‐up image and 500 μm for the whole slice (IBA‐1, microglia, green; Hoechst, nuclei, blue). (c) Microglia density is undisturbed by the IL‐4 challenge at this age, regardless of PFC subregion. Saline/IL‐4 n = 12/12 slices from 6 animals per condition (males are represented as squares and females as diamonds). (d) 3D reconstruction of a microglia with a phagocytic cup in the PFC (IBA‐1, microglia, green; Hoechst, DNA, blue). Scale bar 5 μm. (e) IL‐4 reduces the percentage of microglia with phagocytic cups in a layer and subregion‐dependent manner. IL internal layers are the most affected by IL‐4 challenge. Saline/IL‐4 n = 202–425/144–367 cells from 6 animals per subregion, per layer, per condition. p‐values from a Binomial generalized linear mixed model per subregion per layer with Sidak correction for multiple comparisons. Density data presented as mean ± SEM. *p < 0.05, **p < 0.01 with respect to saline‐injected mice. Cg1, anterior cingulate cortex; IL, infralimbic; MO, medial orbitofrontal cortex; PrL, prelimbic.

In the PFC, microglia were observed to be equally distributed among the different subregions at P10 (Figure 4b,c), suggesting that, contrary to the cerebellum, these cells are likely in a more advanced stage of maturation at this age. IL‐4 administration induced no differences in microglia density in the PFC, independently of the subregion analyzed (Figure 4c). Interestingly, in physiological conditions, the density of microglia in this brain region is, on average, more than double of that in the cerebellum (Figure S2d, LMM, $\widehat{\beta }$ = −0.812, SE = 0.075, t = −10.794, p < 0.001). Nevertheless, IL‐4 induced a significant decrease in the percentage of microglia displaying phagocytic cups (Figure 4d,e), both in layers II/III and V/VI (Figure 4e, B‐GLMM, Layers II/III—$\widehat{\beta }$ = −0.451, SE = 0.258, z = −1.747, p = 0.081; Layers V/VI—$\widehat{\beta }$ = −0.574, SE = 0.235, z = −2.327, p = 0.019), with particular incidence in the layers V/VI of the IL (Figure 4e, B‐GLMM per region with Sidak correction, IL—$\widehat{\beta }$ = −1.409, SE = 0.423, z = −3.333, p = 0.004). Of note, the percentage of microglia with phagocytic cups in the PFC at P10 is much lower compared to the cerebellum, suggesting different programs for microglia‐mediated engulfment of whole cells between these two brain regions.

3.3. Long‐term effects of early IL‐4 administration on microglia of the cerebellum and PFC

Considering the alterations in microglia provoked by IL‐4 exposure at an early age, we hypothesized that this challenge would also affect microglia's phenotype in adult mice. To assess this, we employed a double‐hit model where IL‐4 was injected at P8 and P13 and animals were analyzed at P60 (Figures 5 and 6). Cerebellar microglia showed striking morphological alterations in IL‐4‐injected adult animals, acquiring a hypo‐ramified phenotype characterized by a decrease in occupied territory and number of total ramifications (Figure 5d, LMM for each parameter with Sidak correction, Territory—$\widehat{\beta }$ = −0.234, SE = 0.074, t = −3.135, p = 0.017; Ramifications—$\widehat{\beta }$ = −0.151, SE = 0.048, t = −3.128, p = 0.017), and a Sholl profile with reduced intersections (Figure 5c,e, QS‐GAMM, $\widehat{\beta }$ = −0.079, SE = 0.036, t = −2.199, p = 0.028). At the same time, microglia presented thinner branches, as shown by an increased SA:V ratio (Figure 5d, LMM for each parameter with Sidak correction, SA:V—$\widehat{\beta }$ = 0.227, SE = 0.069, t = 3.236, p = 0.012). Interestingly, the cells that mostly contributed to the observed morphological alterations were restricted to the GL (Figure 5f, QS‐GAMM, $\widehat{\beta }$ = −0.161, SE = 0.056, t = −2.863, p = 0.004), an area where IL‐4 postnatal administration induced a substantial surplus of granule cells in adult mice (Guedes et al., 2023). Cerebellar microglia of IL‐4‐administered animals also displayed a significant increase in total density, in contrast to what was observed at P10 (Figure 5i,j, LMM, $\widehat{\beta }$ = 0.140, SE = 0.051, t = 2.738, p = 0.050). The opposite results of P10 and P60 animals suggest that the IL‐4 postnatal insult induced a (mal)adaptation in microglia's profile that may impair the long‐term function of cerebellar microglia.

FIGURE 5.

Administration of IL‐4 in the second postnatal week induces long‐lasting alterations in cerebellar microglia. (a) IL‐4 injection scheme for assessment of long‐term alterations. (b) Low magnification representative image of the cerebellar region analyzed, subdivided in layers. Scale bar 50 μm. (c) Representative 3D reconstructions of microglia in both conditions (IBA‐1, microglia, green). Scale bar 5 μm. (d, e) Cerebellar microglia of IL‐4‐injected mice display a significantly hypo‐ramified phenotype, occupy a reduced territory and present increased IBA‐1 expression. Sholl‐filled area represents a moving average smoothing of 5 μm intervals. Scale bar for representative traces 15 μm. Saline/IL‐4 n = 43/37 cells from 3 animals per condition. p‐values from a linear mixed model per parameter with Sidak correction for multiple comparisons (c) and a Quasi‐Poisson generalized additive mixed model (d). Territory represents the cell 3D convex hull (μm³); ramifications represent the total number of Sholl intersections; IBA‐1 is calculated as IBA‐1 intensity sum/volume (AU/μm³); SA:V is the ratio between surface area and volume of the cell. (f–h) Sholl analysis shows that microglia of the GL become significantly hypo‐ramified after IL‐4 administration in the second postnatal week (f), while cells from the other layers are unaffected (g, h). Sholl‐filled area represents a moving average smoothing at 5 μm intervals. Scale bar for representative traces 15 μm. GL Saline/IL‐4 n = 18/14; PCL Saline/IL‐4 n = 12/12; ML Saline/IL‐4 n = 14/11 cells from 3 mice per condition. p‐values from a Quasi‐Poisson generalized additive mixed model per layer. (i) Representative images of microglial distribution in the different layers of the cerebellar cortex in both conditions (IBA‐1, microglia, green). Scale bar 15 μm. (j) Postnatal IL‐4 injection leads to an increase in overall microglia density in the cerebellum. Saline/IL‐4 n = 8/9 slices from 3 animals per condition (males are represented as squares and females as diamonds). p‐values from a linear mixed model fit to total density (inset) and per layer with Sidak correction for multiple comparisons. Radar plots and density data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001 with respect to saline‐injected mice. GL, granular layer; ML, molecular layer; PCL, Purkinje cell layer.

FIGURE 6.

Postnatal IL‐4 administration leads to long‐term changes in PFC microglia. (a) Scheme of IL‐4 injection for analysis of P60 PFC microglia. (b) Representative 3D reconstructions of microglia in both conditions. Scale bar 5 μm. (c–f) P60 animals postnatally injected with IL‐4 display no alterations in surface analysis parameters (c) but present significant hypo‐ramification (d), especially in the external layers of the PFC (e). Sholl‐filled area represents a moving average smoothing at 5 μm intervals. Scale bar for representative traces 15 μm. Total Saline/IL‐4 n = 36/26 cells; Layers II/III Saline/IL‐4 n = 17/11 cells; Layers V/VI Saline/IL‐4 n = 19/15 cells from 4 animals per condition. p‐values from a Quasi‐Poisson generalized mixed model fit to the total analysis and per layer. Territory represents the cell 3D convex hull (μm³); ramifications represent the total number of Sholl intersections; IBA‐1 is calculated as IBA‐1 intensity sum/volume (AU/μm³); SA:V is the ratio between surface area and volume of the cell. (g, h) Representative scheme of the analyzed region (g), as well as microglia in PFC layers in both conditions (h) (IBA‐1, microglia, green). Scale bar 15 μm. (i) Postnatal IL‐4 injection increases microglia density specifically in Layers II/III of the IL. Saline/IL‐4 n = 5–15/5–15 slices per region per layer from 5 animals per condition (males are represented as squares and females as diamonds). p‐values from a linear mixed model fit per subregion per layer with Sidak correction for multiple comparisons. Radar plots and density data are presented as mean ± SEM. *p < 0.05, **p < 0.01 with respect to saline‐injected mice. Cg1, anterior cingulate cortex; IL, infralimbic; MO, medial orbitofrontal cortex; PrL, prelimbic.

Although long‐term effects of IL‐4 on PFC microglia were not as pronounced, these cells still presented morphological alterations in adult mice (Figure 6). Despite most surface analysis parameters remaining unchanged (Figure 6c), PFC microglia of animals injected with IL‐4 appear to be significantly hypo‐ramified (Figure 6b,d, QS‐GAMM, $\widehat{\beta }$ = −0.164, SE = 0.067, t = −2.566, p = 0.011), similarly to what was observed in the cerebellum. Microglia of layers II/III accounted for most of the decrease in number of Sholl intersections (Figure 6e, QS‐GAMM, $\widehat{\beta }$ = −0.174, SE = 0.086, t = −2.022, p = 0.043). Furthermore, microglia density was increased specifically in layers II/III of the IL (Figure 6g–i, LMM for each region with Sidak correction, IL—$\widehat{\beta }$ = 0.448, SE = 0.083, t = 5.391, p = 0.003), similarly to what happened in the cerebellar GL. Of note, in adult control mice, microglia of the PFC are significantly different from those of the cerebellum (Figure S4). In the PFC, microglia are more complex and ramified (Figure S4a–c; a—QS‐GAMM, $\widehat{\beta }$ = 0.328, SE = 0.046, t = 7.061, p < 0.001; b—LMM for each parameter with Sidak correction, Ramifications—$\widehat{\beta }$ = 0.524, SE = 0.037, t = 6.489, p = 0.009), present thinner branches and a smaller maximum radius compared to microglia of the cerebellum (Figure S4b, LMM for each parameter with Sidak correction, SA:V—$\widehat{\beta }$ = 0.252, SE = 0.068, t = 3.711, p = 0.011; Max Radius—$\widehat{\beta }$ = −0.473, SE = 0.037, t = −12.697, p < 0.001). In line with these results, cerebellar microglia present a significantly lower solidity, suggesting that a large percentage of the territory occupied by each cell is devoid of ramifications (Figure S4b, LMM for each parameter with Sidak correction, Solidity—$\widehat{\beta }$ = 0.020, SE = 0.001, t = 10.243, p < 0.001). Taken together, our results suggest that the PFC microglial network is tighter due to the higher density of these cells in this brain region (Figure S4d, LMM, $\widehat{\beta }$ = 1.063, SE = 0.034, t = 31.735, p < 0.001) and due to their small and more compact nature. In contrast, microglia of the cerebellum present a looser network characterized by larger, less ramified, and more distant cells, which is in line with previous observations (Stowell et al., 2018).

4. DISCUSSION

The study of the effects of postnatal insults on brain development has been focused on pro‐inflammatory stimuli such as bacterial and viral infections or stress events (Hanamsagar & Bilbo, 2017). These triggers alter the developmental trajectory of neuronal networks and induce phenotypes associated with neurodevelopmental conditions (Baghel et al., 2018; Custódio et al., 2018; Delpech et al., 2016; Li et al., 2018; Reemst et al., 2022). We previously demonstrated that high levels of IL‐4, a type 2 cytokine central to the progression of allergic reactions, induce changes in cerebellar development resulting in hyperactivity and impulsivity in mice (Guedes et al., 2023). In this work, elevated levels of IL‐4 in the periphery in the second postnatal week were induced via administration of house dust mites (model of allergic asthma) or upon intraperitoneal administration of recombinant IL‐4, which mimicked asthma phenotypes (Guedes et al., 2023). Importantly, we showed that IL‐4 acts via microglia, as the abrogation of IL‐4 signaling in these cells rescued microglia‐mediated engulfment of neurons and behavioral alterations upon IL‐4 exposure.

Here, we evaluated the consequences of a premature rise in IL‐4 levels in microglia morphology in the cerebellum and PFC, two brain regions relevant for ASD and ADHD (Curatolo et al., 2010; D'Mello & Stoodley, 2015; Kasparek et al., 2015; Leisman et al., 2023). Despite the difficulty of understanding how specific morphological alterations correlate with microglia function, mounting literature suggests that morphological analysis could be a useful readout of the consequences of different stimuli on microglia phenotypes (Vidal‐Itriago et al., 2022). Microglia phagocytosis of neurons has been associated with morphological alterations, with several studies showing that a reduction in branching and complexity, and the acquisition of amoeboid‐like morphologies, is prevalent in cells engaged in the engulfment process (Chen et al., 2024; Cunningham et al., 2013; Levtova et al., 2017; Morsch et al., 2015). However, this relation is not always straightforward. For example, Perez‐Pouchoulen et al. (2015) studied microglia morphology throughout postnatal development in the rat cerebellum and found that phagocytic cups were exclusively detected in ramified microglia.

In this study, both male and female mice were analyzed. However, we found no significant sex‐dependent differences in any of the parameters. Regardless, there is substantial literature indicating that there are sex‐dependent morphological alterations in microglia in specific circuits, at specific ages, and upon specific insults (Caetano et al., 2017; Nelson et al., 2017; VanRyzin et al., 2019). Despite observing a clear decrease in the percentage of microglia presenting phagocytic cups in IL‐4‐injected mice, suggesting reduced phagocytic activity, we failed to ascertain any morphological alterations connected to this functional output. In the cerebellum, IL‐4 only induced a short‐term alteration in the Sholl profile of microglia of the ML, while most microglia with phagocytic cups are present in the GL, the area of the cerebellar cortex closest to the white matter, from which microglia emerge (Cuadros et al., 1997; Nakayama et al., 2018). This is also the region previously described as presenting the highest density of microglia with phagocytic cups (Perez‐Pouchoulen et al., 2015). Although early‐life IL‐4 exposure also decreased the percentage of microglia with phagocytic cups in the PFC, we also failed to observe short‐term morphological differences. Our data also show that the presence of phagocytic cups as a morphological feature is more likely to be detected in early postnatal developmental periods (Cunningham et al., 2013; VanRyzin et al., 2019), when microglia are maturing toward an adult surveillance phenotype. In fact, we failed to identify these structures in P60 animals, both in the cerebellum and PFC. Still, phagocytic cup structures have been previously identified in adult mice during the clearance of adult‐born neurons in the hippocampus (Kamei & Okabe, 2023) and following injury (Damisah et al., 2020).

In this work, we explored two brain regions that experience extended maturation. The difference in complexity between cerebellar and PFC microglia at P10 indicates that these cells display different developmental programs, which may be related to the different timeframes of neuronal engulfment in these regions. Interestingly, IL‐4 induces an increase in microglia ramification in the ML (the last layer to be populated by microglia) which leads them to attain a Sholl profile close to that of P15 control microglia. These results could indicate that IL‐4 is accelerating the morphological maturation of microglia in this region.

The neurodevelopmental processes occurring at the time of insult may define the susceptibility of each brain region to IL‐4 exposure and the extension of its effects. In terms of neuronal development, in the second postnatal week, Purkinje cells are undergoing dendritic outgrowth and synaptic remodeling (Leto et al., 2016) similarly to what occurs to PFC neurons (Bitzenhofer et al., 2021; Kroon et al., 2019). However, at this age, other major network architectural changes are still taking place in the cerebellum, with the external GL disappearing as granule cells migrate toward the GL (Leto et al., 2016). In parallel, and as occurs in other cortical regions (Lim et al., 2018), the last wave of neuronal death, mostly targeting inhibitory neurons, might be entering its final stage in the PFC. Thus, while in the cerebellum IL‐4 is clearly affecting the development of GCs (Guedes et al., 2023), in the PFC it is likely that this insult is disturbing the end of the process of microglia‐mediated engulfment of inhibitory neurons.

Remarkably, IL‐4 induced similar long‐term consequences on microglia morphology in the cerebellum and PFC. At P60, these cells are characterized by hypo‐ramification and reduced complexity of microglial processes, suggesting that the premature rise of this cytokine in the second postnatal week might influence microglia function in the adult brain, impacting their homeostatic roles and potentially blunting immune reactivity to a second trigger later in life. In fact, similarly to what has been observed for lipopolysaccharide (Cao et al., 2015; Püntener et al., 2012; Wendeln et al., 2018), neonatal IL‐4 overexposure in rats was recently demonstrated to increase the levels of pro‐inflammatory cytokines in adulthood (Wang et al., 2024). Our results are also in agreement with a recent study investigating the effects of early‐life OVA sensitization, followed by allergen challenge (Saitoh et al., 2021), where OVA animals exhibited hypo‐ramified hippocampal microglia, coupled with decreased capacity for synaptic elimination (Saitoh et al., 2021). Interestingly, in the cerebellum, hypo‐ramification of microglia induced by IL‐4 is restricted to the GL, the same region where IL‐4‐injected mice present a surplus of neurons in adulthood (Guedes et al., 2023).

In early life, IL‐4 injection also decreases the total density of cerebellar microglia at P10 in the GL, where most phagocytic processes are taking place. Although IL‐4 has been associated with microglia proliferation (Rossi et al., 2018; Suzumura et al., 1994), our results at P10 suggest that, in this context, IL‐4 might interfere with reduced microglia migration (from the white matter outwards). In contrast to P60, IL‐4‐injected mice show an increased density of microglia in both cerebellum and PFC, which suggests that there is a compensatory mechanism that can be intrinsic to microglia and/or dependent on network activity. One explanation for this finding might be that, despite a slowdown in migration, IL‐4 may also inhibit microglial apoptosis. Alternatively, the increase in neuronal density in the GL might require microglia numbers to adapt and consequently adopt a hypo‐ramified phenotype associated with a smaller surveillance territory. Interestingly, these alterations seem to impact specific subregions within the PFC as well. In IL‐4 mice, microglia density is only significantly increased in layers II/III of the IL. Of note, the IL has previously been shown to control the activity of dopaminergic neurons in the ventral tegmental area (Moreines et al., 2017; Patton et al., 2013), a pathway which can be important in the feedback regulation of dopamine levels in the PFC and in ADHD pathogenesis (Mehta et al., 2019). Furthermore, the disruption of neuronal activity in the IL is known to lead to impulsive behaviors (Sasamori et al., 2019; Tsutsui‐Kimura et al., 2016), a trait displayed by IL‐4‐injected animals (Guedes et al., 2023).

Together, our results show that a premature increase in IL‐4 levels in the postnatal period can modulate microglia phagocytosis across different brain regions without requiring the adoption of amoeboid‐like morphologies or drastically changing their branching patterns. Importantly, the transient presence of this insult is sufficient to elicit long‐lasting microglial network alterations, which might increase their susceptibility to insults later in life and provide an interesting clue towards understanding the link between allergies and neurodegenerative diseases, such as Alzheimer's disease (Joh et al., 2023).

AUTHOR CONTRIBUTIONS

Pedro A. Ferreira: Conceptualization; methodology; software; investigation; statistics; writing – original draft. Carolina Lebre: Conceptualization; methodology; investigation; writing – original draft. Jéssica Costa: Investigation. Francisca Amaral: Investigation. Rosário Ferreira: Investigation. Ana L. Cardoso: Conceptualization; methodology; funding acquisition; supervision. Filipe Martinho: Statistics. Vitor H. Paiva: Statistics; João Peça: Conceptualization; methodology; funding acquisition; supervision. Joana R. Guedes: Conceptualization; methodology; investigation; supervision; funding acquisition; writing – original and final draft.

CONFLICT OF INTEREST STATEMENT

The authors declare they have no conflict of interests.

PEER REVIEW

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/jnc.16266.

Supporting information

Data S1.

Ferreira, P. A. , Lebre, C. , Costa, J. , Amaral, F. , Ferreira, R. , Martinho, F. , Paiva, V. H. , Cardoso, A. L. , Peça, J. , & Guedes, J. R. (2025). Early‐life IL‐4 administration induces long‐term changes in microglia in the cerebellum and prefrontal cortex. Journal of Neurochemistry, 169, e16266. 10.1111/jnc.16266

DATA AVAILABILITY STATEMENT

Data reported in this paper will be shared by the lead author upon request.