Altered synaptic connectivity and brain function in mice lacking microglial adapter protein Iba1

Significance

Abnormal microglia–neuron interaction is increasingly implicated in neurodevelopmental and neuropsychiatric conditions, such as autism spectrum disorders and schizophrenia, as well as in neurodegenerative disorders, such as Alzheimer’s disease. This study demonstrates that the deletion of the microglia-specific protein Iba1, which has long been utilized as a selective microglial marker but whose role has remained unidentified, results in microglial structural and functional impairments that significantly impact synaptic development and behavior. These findings not only highlight the importance of microglia in brain function but may also suggest that modifying the microglial function could provide a therapeutic strategy for the treatment of neurodevelopmental, neuropsychiatric, and neurodegenerative disorders.

Abstract

Growing evidence indicates that microglia impact brain function by regulating synaptic pruning and formation as well as synaptic transmission and plasticity. Iba1 (ionized Ca⁺²-binding adapter protein 1), encoded by the Allograft inflammatory factor 1 (Aif1) gene, is an actin-interacting protein in microglia. Although Iba1 has long been used as a cellular marker for microglia, its functional role remains unknown. Here, we used global, Iba1-deficient (Aif1^−/−) mice to characterize microglial activity, synaptic function, and behavior. Microglial imaging in acute hippocampal slices and fixed tissues from juvenile mice revealed that Aif1^−/− microglia display reductions in ATP-induced motility and ramification, respectively. Biochemical assays further demonstrated that Aif1^−/− brain tissues exhibit an altered expression of microglial-enriched proteins associated with synaptic pruning. Consistent with these changes, juvenile Aif1^−/− mice displayed deficits in the excitatory synapse number and synaptic drive assessed by neuronal labeling and whole-cell patch-clamp recording in acute hippocampal slices. Unexpectedly, microglial synaptic engulfment capacity was diminished in juvenile Aif1^−/− mice. During early postnatal development, when synapse formation is a predominant event in the hippocampus, the excitatory synapse number was still reduced in Aif1^−/− mice. Together, these findings support an overall role of Iba1 in excitatory synaptic growth in juvenile mice. Lastly, postnatal synaptic deficits persisted in adulthood and correlated with significant behavioral changes in adult Aif1^−/− mice, which exhibited impairments in object recognition memory and social interaction. These results suggest that Iba1 critically contributes to microglial activity underlying essential neuroglia developmental processes that may deeply influence behavior.

Microglia are brain macrophages derived from yolk sac progenitors and classically assigned with roles in immune surveillance and response to injury or disease states (1, 2). Recent views, however, indicate that microglia can regulate brain function by remodeling neuronal circuitry both in the healthy and diseased brain (3, 4). For example, excessive complement activity leading to elevated synaptic pruning is implicated in schizophrenia development (5), and cell culture experiments from schizophrenia patient–derived microglia and neurons exhibit an enhanced microglial pruning of dendritic spines (6). A large multilevel study of transcriptional regulation reveals that a subset of microglial-enriched genes is strongly up-regulated in autism spectrum disorders (ASD) peaking during early development, whereas the same module is down-regulated in patients with schizophrenia or bipolar disorder after the age of ∼30 (7). An independent study has further demonstrated alterations in gene expression of specific cell types, including microglia from human patients diagnosed with ASD (8). Thus, understanding the contribution of microglia–neuron interactions to mental health is crucial.

Mice deficient in complement receptor 3 (CR3/CD11b) and fractalkine receptor (CX3CR1) have provided insights into the role of microglia in the synaptic pruning of retinal ganglionic cell axon terminals and of dendritic spines in CA1 hippocampus, respectively (9–11). Furthermore, the loss of synaptic pruning in CA1 hippocampus correlates with behavioral alterations in sociability and repetitive behavior reminiscent of an ASD phenotype in mice (12, 13). Inappropriate microglial activation and enhanced synaptic pruning are also correlated with cognitive deficits in a mouse model of Down syndrome (14). Moreover, microglia perform contact-independent synaptic pruning and potentially affect synaptic transmission through the neuronal activity–driven release of a TNF family cytokine, TWEAK, by microglial cells (15, 16). On the other hand, microglia can contribute to synapse formation in the developing CA1 hippocampus, somatosensory cortex, and in the adult motor cortex, in part by neuronal activity dependent, as well as by basal release of BDNF and IGF1 (17–19). Another mechanism supporting synapse formation implicated extracellular matrix remodeling by microglia (20). Lastly, neuronal activity triggered by animal experience can recruit microglia for structural and functional modifications of neurons (18, 21–25). Despite these advances, specific contributions of microglia in neurodevelopment and circuit remodeling are not clearly understood.

Ionized Ca⁺²-binding adapter protein 1 (Iba1), encoded by the Allograft inflammatory factor 1 (Aif1) gene, is a conserved, intracellular, and Ca⁺²-binding adapter protein of proinflammatory nature, that is selectively expressed by microglia and macrophages and has long been utilized as a microglial marker (26–28). In vitro overexpression studies have shown that Iba1 interacts with the actin cytoskeleton in membrane ruffles and activates Rac GTPase signaling (29–31). Mice lacking Iba1 (Aif1^−/− mice) appear grossly normal, breed well, and show enhanced protection from the development of inflammation (32, 33). In addition, macrophages lacking Iba1 display migration and phagocytosis deficits and secrete reduced levels of proinflammatory cytokines (34). While these in vitro studies identify relevant Iba1 functions for microglia, the in vivo roles of Iba1 in microglia and in normal brain function are largely unknown.

In this study, we specifically investigated whether microglia-lacking Iba1 display any abnormalities and probed for potential alterations in synaptic properties of Iba1-deficient mice. Our findings suggest that Iba1 is required for normal microglial morphology and motility. Importantly, Iba1-deficient mice display strong, persistent synaptic deficits in the CA1 hippocampus, as well as behavioral alterations in object recognition memory and social interaction. Furthermore, our findings support a preferential role for Iba1 in synaptic formation over pruning in juvenile mice. Our study establishes Iba1 as an important modifier of microglia structure and function and microglia–synapse interactions. This study reports on the previously unidentified role of Iba1 protein in the healthy brain.

Results

Iba1 Contributes to Postnatal Microglial Activity.

Previous work showed that Iba1 messenger RNA (mRNA) expression is developmentally regulated, with the highest expression observed during the first two postnatal weeks followed by a drop in young adult stages (17). In addition, an influx or expansion of microglia in the brain occurs during the second postnatal week, particularly in the hippocampus, coinciding with extensive synaptic remodeling (35–37). Together, these observations suggest that microglia play an important role during development. Thus, in this study, we primarily focused on microglial development and function between the second and third postnatal weeks. Accordingly, we first confirmed the absence of Iba1 protein from P16 to P19 (juvenile) Aif1^−/− brain tissues using immunohistochemistry and immunoblotting (SI Appendix, Fig. S1 A and B).

To begin characterizing microglia in Aif1^−/− mice, we first assessed their presence in juvenile Aif1^−/− mice by immunostaining for a microglia-specific protein, P2RY12 (38). We observed that, in various areas of Aif1^−/− brains, microglia were present and maintained at comparable densities as in brains of wild-type mice (Fig. 1 A and B). Next, to determine whether the loss of Iba1 altered the expression of microglial-enriched proteins, we performed immunoblotting for CD11b, TREM2, and CX3CR1 (associated with synaptic pruning) and P2RY12 and MafB (associated with maturation and homeostasis) using juvenile, whole-brain lysates (9–11, 21, 39, 40) (Fig. 1C). We observed that synaptic pruning–associated markers were either up-regulated (CX3CR1 and TREM2) or down-regulated (CD11b) in the Aif1^−/− brain tissues (Fig. 1 C and D). Given that both microglial and peripheral macrophage densities were unchanged in Aif1^−/− mice, the changes in receptor expression likely reflected alterations of protein level per cell in Aif1^−/− mice (Fig. 1) (33). On the other hand, we found no changes in the level of P2RY12 and MafB in Aif1^−/− brains (Fig. 1 C and D). These results suggest that while Iba1 is dispensable for microglial generation, it may be required for microglial function.

Fig. 1.

Juvenile Aif1^−/− mice show normal microglial density but an altered, whole-brain expression of microglial markers of synaptic pruning. (A) Photomicrograph of P16 to P19 fixed frozen brain sections immunostained with an anti-P2RY12 antibody. DG, dentate gyrus and OB, olfactory bulb. (B) Quantitation of P2RY12⁺ microglia in A. n = 10 to 15 fields per region from five mice each genotype, unpaired Student’s t test. (C) Immunoblotting of P16 to P19 whole-brain, whole-cell lysates using antibodies against microglial-enriched synaptic remodeling associated (CD11b, TREM2, and CX3CR1) proteins and homeostatic and maturational (P2RY12 and MafB) proteins. (D) Quantitation of immunoblots in C. n = 4 mice per genotype, unpaired Student’s t test. Average ± SEM *P < 0.05 and **P < 0.01. n.s. = not significant.

Given that CA1 microglia have previously been implicated in developmental synaptic remodeling (11–13, 19), we focused our attention to this hippocampal area. Aif1^−/− microglia in CA1 were less ramified (but not necessarily more amoeboid) when compared with wild-type microglia in morphometric analyses using P2RY12 labeling (Fig. 2 A and B). The microglial morphological deficit associated with Iba1 loss was also observed in other brain regions, such as the neocortex (Fig. 2 A and B). The microglial ability to interact with neurons and to modulate synaptic plasticity depends largely on their dynamic process motility (23–25). Previous studies in cultured macrophages and microglial cell lines using an Aif1 overexpression system suggested that Iba1 localizes with actin cytoskeleton in the membrane ruffles and facilitates motility (29–31). Thus, we investigated whether microglia in acute hippocampal slices display any abnormal motility upon the genetic loss of Aif1. Using two-photon microscopy, we acquired z-stack images of CA1 microglia labeled with Isolectin B4 before and after focal ATP (1 mM) application using a patch-type pipette, as previously described (40–42) (Fig. 2C). We quantified the movement of existing processes within a designated region of interest around the pipette tip (SI Appendix, SI Materials and Methods) and found that microglia in wild-type mice exhibited a greater process motility following ATP application as compared with Aif1^−/− microglia (Fig. 2 C and D).

Fig. 2.

Aif1^−/− microglia in the juvenile brain display reductions in branch complexity and ATP-induced motility. (A) Photomicrograph of P16 to P19 fixed frozen brain sections immunostained with an anti-P2RY12 antibody-facilitated assessment of microglia morphology in hippocampus (Top) and neocortex (Bottom). Sholl analyses were performed by counting the number of branch intersections for each of the concentric dotted circles at 5-μm intervals starting from the soma. (B) Quantitation of microglia branch complexity revealed significant reductions in hippocampus and neocortex of Aif1^−/− microglia. n = 32 to 45 cells from eight to nine fields per region from three mice each genotype. Averages of every intersection were compared between the wild-type and mutant group in a two-way ANOVA, correcting for multiple comparisons using Sidak’s multiple comparison test; hippocampus: F_1,372 = 59.58 and P < 0.0001; neocortex: F_1,492 = 77.15 and P < 0.0001. (C) Representative two-photon images of microglial motility induced by 15 min of ATP application in wild-type (Top) and Aif1^−/− mice (Middle). Microglia were stained with Alexafluor-594–conjugated Isolectin B4 (IB4) in acute hippocampal slices. Note that IB4 also labels blood vessels which have distinct morphological appearances. (Bottom) Artificial cerebrospinal fluid (A-CSF) in the pipette served as a negative control. (D) Quantitation of process motility in a demarcated region of interest (yellow circle in C), indicated a significant deficit in Aif1^−/− microglial motility. Motility index (^+/+: 0.13 ± 0.06; ^−/−: −0.08 ± 0.05; and P = 0.01) was calculated as described in SI Appendix, SI Materials and Methods. n = 18 slices from six mice per genotype, unpaired Student’s t test. Average ± SEM *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. n.s. = not significant.

These results suggest that Iba1 is required for microglial branching complexity and dynamic process motility, implying that Iba1 may contribute to important microglia–neuron interactions in the juvenile brain.

Excitatory Synaptic Connectivity Is Reduced in Aif1^−/− Mice.

Our morphological and biochemical analyses suggested that synaptic remodeling might be affected in juvenile Aif1^−/− mice. To test this possibility, we focused on CA1 pyramidal neurons that have previously been shown to be affected by the genetic manipulation in microglia (11–13, 19). We patch-loaded CA1 pyramidal neurons in juvenile mice with biocytin and performed streptavidin–Alexafluor-647 labeling post hoc to assist confocal microscopy (Fig. 3A). Spine density assessed in stratum radiatum was significantly reduced in Aif1^−/− mice, suggesting a reduction in excitatory synapse number (Fig. 3B). To test whether this structural change was accompanied by a reduction in excitatory drive, we first monitored miniature excitatory postsynaptic currents (mEPSCs) by performing whole-cell patch-clamp of CA1 pyramidal neurons in the presence of tetrodotoxin (0.5 µM) and the GABA_A receptor antagonist, picrotoxin (100 µM) (Fig. 3C). Synaptic events were registered, and both current amplitude and frequency were assessed in wild-type and Aif1^−/− mice (Fig. 3 C and D). While the amplitude of mEPSCs was unaltered, frequency was significantly diminished in Aif1^−/− mice when compared with wild-type mice (Fig. 3D). To determine changes in evoked excitatory transmission, we performed extracellular field recordings of Schaffer collateral–CA1 synapses and found reduced synaptic strength in Aif1^−/− mice as compared to control, as indicated by a reduction of input/output function (SI Appendix, Fig. S2 A and B). In addition, paired-pulse ratio at varying stimulus intervals showed no significant changes, suggesting normal presynaptic function in Aif1^−/− mice (SI Appendix, Fig. S2 C and D).

Fig. 3.

Reduction in excitatory synapses in juvenile Aif1^−/− mice. (A) CA1 pyramidal neurons in P16 to P19 mice were patch loaded with biocytin, fixed, and stained with Alexafluor-647–conjugated streptavidin. Spine images were acquired from apical dendrites within 100 μm distance from soma. (B) Quantitation of spine density (number per 10 µm, ^+/+: 7.93 ± 0.39; ^−/−: 6.47 ± 0.47; and P = 0.02). n = 58 (^+/+) and 46 (^−/−) dendrites from 13 cells from eight mice per genotype, unpaired Student’s t test. (C) Representative traces of whole-cell patch-clamp recording from CA1 pyramidal cells in acute hippocampal slices treated with tetrodotoxin (0.5 µM) and picrotoxin (100 µM). (D) Quantitation of frequency and amplitude of mEPSCs (frequency Hz, ^+/+: 0.27 ± 0.03; ^−/−: 0.16 ± 0.02; and P = 0.02). n = 13 cells from six mice per genotype, unpaired Student’s t test. Average ± SEM *P < 0.05. n.s. = not significant.

To assess whether microglial deficits persist in the adulthood, we performed microglial Sholl and immunoblotting analyses for microglial morphology and synaptic remodeling–associated proteins, respectively, in 3-mo-old wild-type and Aif1^−/− mice. The expression of CD11b, TREM2, and CX3CR1 proteins was normal in 3-mo-old Aif1^−/− brain tissues (SI Appendix, Fig. S3 A and B). In fact, CA1 microglial morphology in adult Aif1^−/− mice was more complex compared to wild-type microglia (SI Appendix, Fig. S3 C and D), suggesting that compensatory changes have occurred postdevelopment to normalize some aspects of microglial activity in Aif1^−/− brains. We further assessed synaptic structure–function, as described (Fig. 3), but in 3-mo-old wild-type and Aif1^−/− mice (Fig. 4). Intriguingly, we observed a reduction in both the number of dendritic spine (Fig. 4 A and B) and mEPSC frequency (Fig. 4 C and D) in adult Aif1^−/− mice when compared with wild-type mice. In contrast, long-term potentiation (LTP) in the CA1 area was indistinguishable between adult wild-type and Aif1^−/− mice (SI Appendix, Fig. S4).

Fig. 4.

Deficit in excitatory synapse density persists in adult Aif1^−/− mice. (A) CA1 pyramidal neurons in P85 to P90 mice were patch loaded with biocytin, fixed, and stained with Alexafluor-647–conjugated streptavidin. Spine images were acquired from apical dendrites within 100 μm distance from soma. (B) Quantitation of spine density (number per 10 μm, ^+/+: 12.25 ± 0.41; ^−/−: 10.18 ± 41; and P = 0.0007). n = 36 (^+/+) and 32 (^−/−) dendrites from seven to eight cells from four mice per genotype, unpaired Student’s t test. (C) Representative traces of whole-cell patch-clamp recording from CA1 pyramidal cells in acute hippocampal slices treated with tetrodotoxin (0.5 μM) and picrotoxin (100 μM). (D) Quantitation of frequency and amplitude of mEPSCs (frequency Hz, ^+/+: 0.37 ± 0.02; ^−/−: 0.25 ± 0.04; and P = 0.02). n = 7 to 8 cells from four mice per genotype, unpaired Student’s t test. Average ± SEM *P < 0.05 and ***P < 0.001. n.s. = not significant.

Taken together, our results indicate that the excitatory synaptic drive was significantly reduced in juvenile Aif1^−/− mice, at least in part as a result of a decrease in the number of synaptic contacts, and that such alterations persisted in adult Aif1^−/− mice, despite that microglial deficits were largely compensated postdevelopment.

Adult Aif1^−/− Mice Display Behavioral Alterations.

Microglial dysfunction has been correlated with alterations in mnemonic processes and social behavior in mice (12, 13, 43). Based on our synaptic physiology analyses involving CA1 hippocampus, we anticipated that relevant behavior might be affected in adult Aif1^−/− mice. Since Aif1^−/− mice have not been formally characterized in behavioral terms, we first assessed the general locomotion/exploration and anxiety using the open-field test (44, 45) (SI Appendix, Fig. S5A). Compared with wild-type mice, Aif1^−/− mice traveled a greater distance (by 19%) in the open field, suggesting a modest increase in exploration/locomotion in the latter (SI Appendix, Fig. S5A). We further subjected Aif1^−/− mice to the elevated plus maze test, as a definitive measure of anxiety-like behavior (46, 47) (SI Appendix, Fig. S5B). Aif1^−/− and wild-type mice spent an equivalent amount of time in the open arm, indicating no differences in anxiety (SI Appendix, Fig. S5B). Furthermore, Aif1^−/− mice did not preferentially avoid the center area during the open-field test, consistent with an absence of anxiety-like behavior in these mice (SI Appendix, Fig. S5A). Together, these studies show that Aif1^−/− mice display an increase in general exploration/locomotion but lack anxiety-like behavior.

We next performed the novel object recognition test, a task dependent in part on CA1 hippocampus (48–50). In this test, a mouse was first exposed to two identical objects in an open field and then, after an hour of retention, reexposed to the open field with one of the objects switched to a new object (51) (Fig. 5A). Wild-type mice displayed preference for the novel object over the familiar one upon reexposure, as assessed by the discrimination index, whereas such preference was reduced (by 64%) in Aif1^−/− mice (Fig. 5A). Total overall exploration for the novel and familiar object combined remained comparable between genotypes (Fig. 5A). These results suggest that adult Aif1^−/− mice display a strong reduction in object recognition memory, consistent with a persistent deficit in hippocampal synaptic connectivity.

Fig. 5.

Adult Aif1^−/− mice display behavioral alterations. (A) Mice explored an open field guided by visual cues; first in the presence of two identical objects (3 min of training) and an hour later in the presence of two objects, when one of the two old objects was replaced with a new object (3 min of testing). n = 15 (^+/+) and 13 (^−/−). (Left) The percentage exploration for each object during testing was calculated, and statistical significance across different conditions and genotypes was assessed by a two-way ANOVA, correcting for multiple comparisons using Tukey’s multiple comparison test. (Middle) Discrimination index during testing was calculated (^+/+: 0.25 ± 0.04; ^−/−: 0.09 ± 0.04; and P = 0.01) as described in SI Appendix, SI Materials and Methods. (Right) Total exploration during testing (^+/+: 19.88 ± 1.96 s; ^−/−: 16.89 ± 1.54 s; and P = 0.23) was recorded. The statistical significance between two genotypes was assessed using unpaired Student’s t test. (B) Mice explored a three-chamber Plexiglas with sliding doors; first, alone in the center chamber (5 min) with doors closed, then with a stranger mouse of same sex, which was placed in a circular cage in one of the two chambers (10 min) with both doors open (for sociability). n = 23 (^+/+) and 28 (^−/−). (Left) Percentage sniff time was calculated, and statistical significance across different conditions and genotypes was assessed by a two-way ANOVA, correcting for multiple comparisons using Tukey’s multiple comparison test. (Middle) Sociability index was calculated (^+/+: 0.23 ± 0.07; ^−/−: 0.03 ± 0.05; and P = 0.03) as described in SI Appendix, SI Materials and Methods. (Right) Total sniff time during each trial (^+/+: 98.45 ± 5.1 s; ^−/−: 99.45 ± 3.62 s; and P = 0.87) was recorded. The statistical significance between two genotypes was assessed using unpaired Student’s t test. Mice were between 3 to 5 mo. Average ± SEM *P < 0.05 and ****P < 0.0001. n.s. = not significant.

To further investigate potential alterations in neural circuits outside the hippocampus, such as cortical or midbrain structures in Aif1^−/− mice, we performed a three-chamber social test (52–54). A test mouse was assessed for sociability by exposing it to a stranger mouse. Since Aif1^−/− mice had an increase in exploration/locomotion in the open-field test (SI Appendix, Fig. S5A), we assessed sniff time rather than time spent in each chamber as a measure of interaction. Wild-type mice preferentially sniffed the stranger mice more than the empty chamber, while Aif1^−/− mice sniffed both the stranger mice and the empty chamber equally (Fig. 5B). The sociability index was reduced by 87% in Aif1^−/− mice compared with wild-type mice (Fig. 5B). Both groups displayed equivalent total sniff time (Fig. 5B). These results suggest that Aif1^−/− mice exhibit a strong deficit in social interaction, and neural circuit impairment may occur in additional Aif1^−/− brain regions.

In summary, adult Aif1^−/− mice showed significant alterations in behavior that are often associated with neurodevelopmental disorders involving multiple brain areas.

Iba1 Facilitates the Growth of Excitatory Synapses in the Juvenile Brain.

The effect of Iba1 loss on synaptic remodeling function supported the idea that Iba1 promotes the growth of excitatory postsynaptic structures (Figs. 3 and 4). We thus investigated whether the synaptic growth-promoting function of Iba1 could be attributed to an ability to limit synaptic pruning or to facilitate synaptic formation. To assess a role for Iba1 in synaptic pruning, we performed a synaptic engulfment assay using immunostaining and quantitative Airyscan microscopy using P16 to P19 fixed frozen tissue sections from wild-type and Aif1^−/− mice (12, 21, 55). P2RY12, CD68, and PSD-95 served as respective markers for microglia, microglial lysosome, and excitatory postsynaptic structures. Consistent with earlier observations (Fig. 3 A and B), we detected a 50% reduction in the volume of overall PSD-95+ structures in hippocampal CA1 fields of Aif1^−/− mice compared with wild-type mice (Fig. 6 A and B). Airyscan microscopy followed by quantitative analyses using Imaris software revealed no significant change in the volume of CD68+ structures per Aif1^−/− microglial volume (Fig. 6 A and B). Using the “mask function” of Imaris (SI Appendix, SI Materials and Methods), a 45% reduction in the volume of PSD-95+ structures that were associated with microglial CD68+ structures was observed in Aif1^−/− mice as compared to wild-type mice (Fig. 6 C and D related to Movies S1 and S2, and SI Appendix, Fig. S6 related to Fig. 6A). These results suggest that the reduction in the number of excitatory synapses upon Iba1 loss is not due to an enhanced contact-dependent synaptic pruning. While an increase in microglia-driven, contact-independent pruning, which requires sensory experience (15, 16), could potentially contribute to the reduction in number of excitatory synapses in Aif1^−/− mice, it is an unlikely scenario given that our synaptic structure–function analyses were performed under basal conditions (Figs. 3, 4, and 6).

Fig. 6.

Iba1 loss attenuates microglial synapse engulfment in the juvenile mice but also causes an early synapse formation deficit. (A) Confocal z-stack images (maximum projection) of CA1 excitatory synapses (Top) and microglia with lysosomal structures (Bottom) in fixed brain tissues from P16 to P19 mice. (B, Bottom) A reduction in total volume of PSD-95+ structures per CA1 field (^+/+: 0.02 ± 0.001; ^−/−: 0.01 ± 0.001; and P < 0.0001) in Aif1^−/− mice. n = 28 fields from four mice per genotype. A similar volume of CD68+ lysosomal structures (green, white arrows) per microglial (red) volume (^+/+: 0.12 ± 0.01; ^−/−: 0.14 ± 0.01; and P = 0.5) in wild-type and Aif1^−/− mice. n, nucleus. (C) Three-dimensional surface rendering and using “mask” function in Imaris demonstrating postsynaptic structures (PSD-95, blue, black arrows) in microglial (P2RY12, red) lysosomal structure (CD68, green). Zoomed images demonstrate examples of engulfed synaptic materials (a different sets of microglia; Movies S1 and S2). (D, Middle) A reduced percentage volume of engulfed synaptic material (PSD-95+ structures within CD68+ structures) per microglia volume (^+/+: 0.99 ± 0.17; ^−/−: 0.54 ± 0.12; and P = 0.03) in Aif1^−/− mice. n = 24 cells from six mice per genotype. (E) CA1 pyramidal neurons in P8 to P10 mice were patch-loaded with biocytin, fixed, and stained with Alexafluor-594–conjugated streptavidin. Confocal spine images (arrows) were acquired from apical dendrites within 50 μm distance from soma. (F) Quantitation of spine density in P8 to P10 mice (number per 10 µm, ^+/+: 4.00 ± 0.22; ^−/−: 2.66 ± 0.29; and P < 0.0001). n = 41 (^+/+) and 36 (^−/−) dendrites from 10 cells from seven mice per genotype, unpaired Student’s t test. Average ± SEM *P < 0.05 and ****P < 0.0001. n.s. = not significant.

Recent studies show that microglia can contribute to excitatory synapse formation in juvenile mice (18, 19). Iba1 could promote the synapse formation of cortical neurons during early postnatal development (P8 to P10) (17). To determine if Iba1 may perform a similar function in CA1 pyramidal neurons, we further assessed dendritic spine density in P8 to P10 mice at the time when synaptic pruning events are largely absent (11, 19) (Fig. 6E). Our results revealed a 43% reduction in spine density in P8 to P10 Aif1^−/− mice compared with wild-type mice (Fig. 6F). This result suggests that Iba1 deletion impairs synapse formation.

Lastly, astrocytes can contribute to synaptic formation, pruning, transmission, and plasticity, and recent studies position astrocytes downstream of microglia (56–60). We therefore performed density and morphological analyses of CA1 astrocytes using GFAP immunostaining. The assessment of CA1 astrocytic number and morphology failed to reveal significant differences between wild-type and Aif1^−/− mice (SI Appendix, Fig. S7). While it is plausible that astrocytes could still affect synaptic processes despite their normal number and morphology in the absence of Iba1, this possibility remains to be fully investigated.

Together, our results suggest that Iba1 contributes to an overall growth of excitatory synapses by virtue of its predominant role in synapse formation over pruning in the juvenile brain.

Discussion

We report that mice lacking Iba1 have altered microglial structure–function, deficits in excitatory synapses, and changes in behavior. Our findings highlight a pivotal contribution of microglial activity during neurodevelopment to synaptic refinement that may influence behavior. This study provides a characterization of Iba1 contributions to microglial biology and brain function using a genetic loss-of-function approach.

Microglia are a selective cell population in brain tissue that can migrate, dynamically extend, or retract processes either to probe local environments (surveillance) or to respond to pathogens or injury (chemotaxis) (41, 61). In mice with the genetic inactivation of Iba1 expression, we identified a significant impairment in microglial ramification and ATP-induced motility in acute brain slices, suggesting that Iba1 may indeed play essential roles in the dynamic motility of these cells (Fig. 2). Microglial motility and associated neuronal plasticity are engaged by the ATP-driven activation of microglial purinergic receptor P2RY12 (23, 25, 41, 42). However, we did not find changes in P2RY12 protein level in Aif1^−/− brain tissue, suggesting that P2RY12 may act upstream of Iba1 to enable motility (Figs. 1 and 2). In addition, focal laser injury can enhance Ca⁺² activity in microglial soma and protruding processes (62). Although its role in intracellular Ca⁺² dynamics has not been investigated, Iba1 protein is known to bind Ca⁺² and to activate Rac GTPase signaling (31, 63)—these Iba1 functions could potentially contribute to microglial process extensions that may deeply influence neuroglia interactions.

Recent studies demonstrate that microglia are not only essential for synaptic pruning but may also participate in synapse formation during early neurodevelopmental stages (17–19). In the somatosensory cortex, high-Iba1 mRNA levels have been correlated with synapse formation in early postnatal mice (17). Consistent with this observation, our neuronal structural analysis revealed strong reductions of excitatory synaptic density in the CA1 hippocampus of early postnatal as well as juvenile Aif1^−/− mice (Figs. 3 and 6). Paradoxically, we also observed a reduction in microglial synaptic engulfment capacity in hippocampal CA1 when Iba1 is absent (Fig. 6 C and D and SI Appendix, Fig. S6). Given that synaptic formation precedes pruning and pruning peaks between the second through third postnatal week with concomitant synaptic formation, it is conceivable that, of the two remodeling processes, synaptic formation is a more dominant process in juvenile mice (11, 12, 18, 64). Consistent with a role of Iba1 in synapse formation, Aif1^−/− synaptic deficits were apparent by P8 to P10, when synaptogenesis reaches its peak in somatosensory cortex and to a large extent in the hippocampus (17, 65). It is also possible that reduction in the capacity of microglial synaptic engulfment in juvenile Aif1^−/− mice could be in part due to compensatory changes to an early synapse formation deficit in these mice. Future approaches should consider the temporal ablation of Iba1 function to better study the role of microglia in defining the “critical period plasticity” (36, 37).

Our study suggests that Iba1 can impact object recognition memory, consistent with its roles in CA1 excitatory synapse structure–function in adulthood (Figs. 4 and 5A). We, however, did not observe a significant difference in CA1-LTP between adult wild-type and Aif1^−/− mice (SI Appendix, Fig. S4). Object recognition memory depends on the CA1 area only in part (48–50), and whether circuits in other brain areas display altered synaptic structure–function and LTP impairments in adult Aif1^−/− mice remain to be investigated.

How exactly Iba1 regulates microglial function warrants further investigation and a deeper mechanistic understanding of the cellular role of Iba1. Although generally described as a cytosolic protein, Iba1 subcellular localization and function remains to be determined. Iba1 activities within cells may include, but are not limited to, 1) actin-based cytoskeletal roles that drive microglial phagocytotic and/or secretory functions, such as vesicular exocytosis or the exosomal release of cytokines and microRNAs (66) and 2) the regulation of microglial expression of genes encoding critical receptors, cytokines, and matrix metalloproteases (21, 67). We found the decreased ramification and motility of microglia-lacking Iba1, consistent with altered cytoskeletal functions (Fig. 2). We also identified clear differences in protein levels associated with the synaptic pruning in Aif1^−/− brain tissues (Fig. 1), but our observations so far cannot distinguish between possible direct or indirect mechanisms by which Iba1 could regulate specific microglial gene expression during neurodevelopment.

Human studies demonstrate that enhanced Iba1 expression is correlated with the onset of ASD and Down syndrome during early life (7, 14). In addition, Iba1 is down-regulated with the onset of schizophrenia and bipolar disorder around age 30 (7). These results are consistent with microglial roles in neurodevelopment and neurophysiology and the idea that their dysfunction correlates with onset of neurodevelopmental and neuropsychiatric disorders. Human epidemiological studies have clearly demonstrated a link between early immune activation and psychiatric conditions in later life, including depression and psychosis (68, 69). Furthermore, prenatal infections have been linked to the development of ASD and schizophrenia (70, 71). Our Iba1 loss-of-function model of microglial dysfunction supports the impact of early developmental changes to the alterations of cognitive and social behaviors in adulthood (Fig. 4). These results reinforce the need to deeply assess microglial properties during those early immune perturbations leading up to the development of cognitive dysfunctions later in life. Recent work has demonstrated that combined auditory and visual stimulation to induce gamma oscillations can activate microglia and lead to reductions in amyloid plaques that are associated with cognitive improvements (72). Furthermore, the microglial pruning of engram cell synapses correlates with the forgetting mechanism in adult mice (43). As microglia function becomes further implicated in neurodevelopmental, neuropsychiatric, and neurodegenerative disorders, a better understanding of neuroglia interactions will provide therapeutic insights.

Materials and Methods

Aif1^−/− mice were generated, maintained, and bred as previously described (32, 33) and were further backcrossed to C57BL/6J for more than 10 generations (34). All experimental procedures involving mice were carried out following the guidelines of the institutional animal care and use committee (IACUC) at the Albert Einstein College of Medicine. Information on antibodies and procedures involving immunohistochemistry, hippocampal slice preparation, electrophysiology, behavior, microglial morphology and motility, and the biochemical evaluation of brain tissues were detailed in SI Appendix, SI Materials and Methods. In all cases, the investigators were blind to the conditions and genotypes during data acquisition and analysis.

Data Availability

All study data are included in the article and/or supporting information.