Elsevier

Neuroscience Research

Volume 100, November 2015, Pages 1-5
Neuroscience Research

Review article
Microglia in the pathogenesis of autism spectrum disorders

https://doi.org/10.1016/j.neures.2015.06.005Get rights and content

Highlights

  • We discussed the possible involvement of microglia in the pathogenesis of ASD.

  • Impairments in synapse E/I balance in ASD are discussed.

  • The characteristics of microglia in ASD are discussed.

  • The role of the hippocampus in ASD is discussed.

Abstract

Proper synaptic pruning is essential for the development of functional neural circuits. Impairments in synaptic pruning disrupt the excitatory versus inhibitory balance (E/I balance) of synapses, which may cause neurodevelopmental disorders such as autism spectrum disorder (ASD). Recent studies have determined molecular mechanisms by which microglia, the brain's resident immune cells, engulf inappropriate and less active synapses. Thus, microglial dysfunction may be involved in the pathogenesis of ASD through attenuated or excess synaptic pruning. In this review, we discuss recent animal and human studies that report an E/I imbalance and the characteristics of microglia in ASD. We will further discuss whether and how synaptic pruning by microglia is involved in the pathogenesis of ASD.

Introduction

Microglia are resident immune cells with the phagocytic capacity to engulf dead cells and cell debris during inflammation in the brain. Microglia had long been believed to derive from peripheral macrophages that enter the brain after birth. However, it is now known that microglia develop from erythromyeloid progenitors in the early embryonic yolk sac and migrate into the brain, where they reside for life, most likely sustaining the microglial population. Recent studies have demonstrated that microglia play an important role not only during inflammation but also in the healthy brain. For example, microglia preferentially engulf weak or less active synapses, contributing to the development of refined functional neural circuits with strong or more active synapses (Schafer et al., 2012, Paolicelli et al., 2011).

In mouse models of obsessive–compulsive disorder (Chen et al., 2010) and Rett syndrome (Derecki et al., 2012), disorders that exhibit behavioral phenotypes similar in part to autism spectrum disorder (ASD), deficits in synaptic pruning have been reported. Both attenuated and excess synaptic pruning by microglia could impair the brain's excitatory versus inhibitory (E/I) balance. An impairment in synapse E/I balance has been suggested as a common mechanism underlying several neurodevelopmental disorders, including ASD, epilepsy, Rett syndrome, and Fragile X syndrome (Gatto and Broadie, 2010). Furthermore, the risk genes for ASD include a number of synapse-related genes, such as members of the neuroligin and neurexin families, synaptic adhesion molecules whose expression is significantly upregulated during the perinatal period, suggesting that defects in synapse formation are likely involved in the pathogenesis of ASD (State and Šestan, 2012).

Recently, interesting evidence linking reduced synaptic pruning during development to ASD has been reported (Tang et al., 2014). The authors found that dendritic spine density was increased in postmortem ASD specimens in layer V pyramidal neurons in the superior middle temporal lobe, Brodmann Area 21 – a region implicated in ASD due to its participation in the brain networks involved in social interaction and communication. The authors categorized ASD patients by age into 2–9 years old (children) and 13–19 years old (adolescents) and found that the spine density in the childhood ASD group was comparable to that of an age-matched control group, while the spine density in the adolescent ASD group was significantly higher than that in the age-matched control group. These results suggested that the developmental pruning of excitatory synapses was attenuated in ASD patients; however, the involvement of microglia was not investigated in this study. It is of interest to determine whether and how impairments in synaptic pruning are a primary cause of ASD and whether attenuated or enhanced synapse engulfment by microglia is involved in this process.

Neural circuits mature through synapse formation, pruning, and maturation. These synaptogenic processes are fundamental for the development of proper E/I balance in the brain. In ASD patients, the disruption of synapse E/I balance, especially an increased ratio of E/I, has been suggested as a leading cause for classical hallmark phenotypes in ASD, including fundamental impairments in social interaction and language communication accompanied by highly restrictive interests and/or repetitive behaviors (Rubenstein and Merzenich, 2003).

The increased excitation theory is partly supported by a high comorbidity (10–30%) between ASD and epilepsy that is caused by synchronized neuronal hyperactivity (Canitano, 2007, Tuchman et al., 2010). Experimentally, optogenetic activation of excitatory neurons in the mouse medial prefrontal cortex, a brain region important for regulating executive functions in social interaction, resulted in impairments in social behaviors (Yizhar et al., 2011). In contrast, the activation of inhibitory neurons in the medial prefrontal cortex did not affect social behaviors. Mice lacking IRSp53 (insulin receptor substrate protein, 53 kDa), which is an excitatory synaptic signaling scaffold protein and is also known as an ASD risk gene, displayed enhanced NMDA receptor function in the hippocampus and impaired social interaction and ultrasonic vocalization-mediated communication (Chung et al., 2015). The social deficits in the IRSp53−/− mice were rescued by an intraperitoneal injection of the NMDA antagonist memantine and by the metabotropic glutamate receptor 5 (mGluR5) antagonist MPEP.

Before discussing the possible involvement of microglia in synapse E/I balance in ASD, we will introduce clinical findings on the characteristics of microglia in ASD patients. Accumulating evidence suggests that microglia are activated in various brain regions, including the prefrontal cortex, in some ASD patients. Suzuki and colleagues conducted a positron emission tomography (PET) analysis of young adult ASD patients (age range 18–31 years) using a radiotracer that specifically binds to activated microglia (Suzuki et al., 2013). The authors defined regional brain radiotracer binding potential as representative of microglia activation and found microglial activation in brain regions such as the anterior cingulate and the cerebellum, whose dysfunction has been suggested in ASD. Activation of microglia in living ASD patients (age range 3–10 years) has also been suggested by the marked elevation of proinflammatory chemokines such as MCP-1 (Monocyte Chemoattractant Protein-1), which is released from activated microglia, in cerebrospinal fluid (Vargas et al., 2005).

In addition, several postmortem studies have suggested that the anatomical and morphological characteristics of microglia are abnormal in the brains of individuals with ASD. The density of microglia in gray matter but not in white matter of the dorsolateral prefrontal cortex has been shown to significantly increase in ASD patients compared to control subjects (Morgan et al., 2010). It is notable that the trend increased with age, which raises the possibility that microglial activation is a symptom but not a cause of ASD. Tetreault and others found that the density of microglia increased significantly in two disparate cortical areas (i.e., fronto-insular and visual cortex) in individuals with ASD versus controls (Tetreault et al., 2012), concluding that microglia are present at a higher density in ASD patients throughout the cerebral and cerebellar cotices in combination with the reports showing the increased density of microglia in the dorsal lateral prefrontal cortex (Morgan et al., 2010) and the cerebellum (Vargas et al., 2005).

The morphology of microglia has also been assessed in postmortem studies. In 5 of 13 ASD patients, microglia appeared markedly activated in the dorsolateral prefrontal cortex accompanied by somal enlargement, process retraction and thickening, and filopodia extension from processes, all indicative of activated microglia in the immune response (Morgan et al., 2010). It is noteworthy that the microglial abnormalities were observed in not all but some of the ASD patients. It is awaited to determine the phenotypic characteristics of individuals with ASD who have abnormal microglia in the brain. An increased interaction between microglia and neurons has also been suggested in the dorsolateral prefrontal cortex; microglia are more frequently found near neurons, with microglial processes encircling neurons, in ASD patients (Morgan et al., 2012). The authors noted that the increased microglia–neuron interactions were present from an early age in ASD patients by analysis of a young subject subgroup, which may indicate that an altered microglia–neuron interaction is a possible cause of ASD.

It is intriguing to clarify whether the microglial abnormalities, including density, function, and morphology, in the restricted brain regions in each individual with ASD are associated with the phenotypic heterogeneity of ASD. However, to our knowledge, no reports have ever investigated these points in detail. Because it has been suggested that the long-distance functional connectivity between brain regions is weakened in ASD (Belmonte et al., 2004, Courchesne and Pierce, 2005, Schipul et al., 2011), the microglial abnormalities and resulting impairments in synaptic pruning in any brain regions may induce differences in the behavioral phenotypes of individuals with ASD.

Recent studies have clarified cellular and molecular mechanisms that regulate synaptic pruning by microglia in the healthy developing brain: complement-mediated synaptic pruning by microglia in the visual thalamus (Schafer et al., 2012) and microglia-dependent synapse elimination in the hippocampus (Paolicelli et al., 2011). In the following section, we will discuss the possible involvement of complement proteins and microglia in the pathogenesis of ASD (Fig. 1).

In the developing retinogeniculate system of binocular animals, dorsal lateral geniculate nucleus (LGN) relay neurons are initially innervated by multiple RGC (retinal ganglion cell) axons. Gradually, over the first month of development, the excess or inappropriate synapses are pruned, and the correct synapses are strengthened in an activity-dependent manner (Campbell and Shatz, 1992, Huberman et al., 2008, Sretavan and Shatz, 1986). One molecular pathway indicated in activity-dependent synaptic pruning by microglia in the dorsal LGN is the classical complement cascade (Schafer et al., 2012, Stevens et al., 2007). The complement protein C1q, the initiating protein in this pathway, and the downstream complement protein C3 opsonize or tag unwanted cells or debris for removal by phagocytic macrophages via specific complement receptors. These complement proteins are highly localized to immature synapses and are required for synaptic pruning in the retinogeniculate system (Stevens et al., 2007). Further, it was suggested that ‘weaker’ or less active RGC synapses are tagged by the C3 protein and then detected and pruned by microglia that express complement receptor 3 (CR3/cd11b) (Schafer et al., 2012). C1q expression was increased in the peripheral serum of children with ASD (Corbett et al., 2007), leading us to hypothesize that elevated levels of complement proteins in the ASD brain promote synaptic pruning by microglia. Alternatively, ubiquitously increased complement proteins may cover the specific localization of complement proteins on less active synapses, which may prevent the proper pruning of inappropriate synapses by microglia. Future studies are required to clarify whether and how complement protein–microglia interactions impair synapse E/I balance.

The role of the hippocampus in ASD is beginning to receive notice. In the hippocampal CA1 pyramidal neurons of BTBR mice, a model of idiopathic autism, the frequency of spontaneous inhibitory postsynaptic currents (IPSCs) was significantly reduced, while the amplitude and frequency of spontaneous excitatory postsynaptic currents (EPSCs) were substantially increased compared with those in control mice (Han et al., 2014). In another possible model of ASD, rats exposed in utero to valproate, the frequency of spontaneous EPSCs was increased in CA3 pyramidal neurons when compared with control rats (Tyzio et al., 2014). When the phosphatase Pten (Phosphatase and tensin homolog on chromosome ten), which is mutated in some ASD patients with macrocephaly (Conti et al., 2012), was knocked out in dentate granule cells, these neurons exhibited hyperactivity accompanied by an increased number of dendritic spines and excitatory synapses (Williams et al., 2015). Finally, rats whose ventral hippocampus was lesioned with ibotenic acid at 7 days old showed impaired social behaviors when tested at the age of 13 weeks (Becker et al., 1999).

In the CA1 region of the postnatal hippocampus (2–3 weeks) in mice lacking the fractalkine receptor CX3CR1, the density of microglia was reduced, which resulted in impaired pruning of excitatory post-synapses on CA1 pyramidal neurons. As a result, the synaptic maturation of CA1 pyramidal neurons, as indicated by an increase in the frequency of sEPSCs and the sEPSC/miniature EPSC ratio, which reflects increased connectivity of afferent synaptic inputs, was impaired (Paolicelli et al., 2011). Fractalkine, or CX3CL1, is a chemokine produced by neurons and binds to the CX3CR1 receptor, which is selectively expressed on the surface of microglia (Ransohoff, 2009) and mediates neuron–microglia interactions (Paolicelli et al., 2014, Wolf et al., 2013). In CX3CR1 knockout mice, functional magnetic resonance imaging (fMRI) analysis revealed that functional connectivity between the hippocampus and prefrontal cortex was impaired (Zhan et al., 2014). Importantly, these mice showed decreased social interactions and increased repetitive behaviors, phenotypes associated with ASD. These results suggest that impaired synaptic pruning by microglia in the hippocampus contributes to the development of ASD. However, it should be noted that synaptic pruning may be impaired in the brain regions other than the hippocampus in CX3CR1 knockout mice as it is likely that all microglia in the brain express CX3CR1 (Harrison et al., 1998, Jung et al., 2000).

Our group is examining the possible role of the dentate gyrus subregion of the hippocampus in ASD primarily for the following two reasons. First, the dentate gyrus is suggested to regulate pattern separation, a process by which similar inputs are transformed into less similar outputs, and pattern completion, a process by which complete stored representations are reconstructed from partial inputs that are part of the stored representation (Colgin et al., 2008, Sahay et al., 2011). Thus, we hypothesize that deficits in pattern separation and/or pattern completion may result in repetitive and restricted behaviors observed in individuals with ASD. Second, the dentate gyrus is one of the rare brain regions where continuous neurogenesis occurs throughout life. This property would provide us with the opportunity to increase the activity of adult-born neurons to compete with pre-existing immature synapses formed by postnatal-born neurons in ASD. The roles of microglia in the development and function of dentate gyrus-hippocampal CA3 synapses in ASD is currently being clarified in our laboratory.

Section snippets

Conclusions

Whether and how synaptic pruning by microglia contributes to the pathogenesis and pathophysiology of ASD remains elusive. Analysis of postmortem specimens from individuals with ASD have suggested that microglia are activated in the ASD brain; however, it remains unclear whether microglia are activated or inactivated in the early pathogenic process of ASD during postnatal periods when impaired synaptic pruning may take place.

An association between maternal immune activation (MIA) and ASD in

Acknowledgement

This work was supported by JSPS KAKENHI Grant Numbers 26460094 and 26117504.

References (42)

  • B. Stevens et al.

    The classical complement cascade mediates CNS synapse elimination

    Cell

    (2007)
  • G. Tang et al.

    Loss of mTOR-dependent macroautophagy causes autistic-like synaptic pruning deficits

    Neuron

    (2014)
  • R. Tuchman et al.

    Autism and epilepsy: historical perspective

    Brain Dev.

    (2010)
  • A. Becker et al.

    Social behaviour in rats lesioned with ibotenic acid in the hippocampus: quantitative and qualitative analysis

    Psychopharmacology (Berl)

    (1999)
  • M.K. Belmonte et al.

    Autism and abnormal development of brain connectivity

    J. Neurosci.

    (2004)
  • G. Campbell et al.

    Synapses formed by identified retinogeniculate axons during the segregation of eye input

    J. Neurosci.

    (1992)
  • R. Canitano

    Epilepsy in autism spectrum disorders

    Eur. Child Adolesc. Psychiatry

    (2007)
  • W. Chung et al.

    Social deficits in IRSp53 mutant mice improved by NMDAR and mGluR5 suppression

    Nat. Neurosci.

    (2015)
  • S. Conti et al.

    Phosphatase and tensin homolog (PTEN) gene mutations and autism: literature review and a case report of a patient with Cowden syndrome, autistic disorder, and epilepsy

    J. Child Neurol.

    (2012)
  • B.A. Corbett et al.

    A proteomic study of serum from children with autism showing differential expression of apolipoproteins and complement proteins

    Mol. Psychiatry

    (2007)
  • N.C. Derecki et al.

    Wild-type microglia arrest pathology in a mouse model of Rett syndrome

    Nature

    (2012)
  • Cited by (0)

    View full text