Keywords

Autism is a pervasive, debilitating, neurodevelopmental disorder characterized by three core symptoms: (1) abnormal social interaction; (2) impaired verbal and nonverbal communication; and (3) the presence of restrictive, repetitive or stereotyped behaviors (APA 2013). Currently, the prevalence of autism is rising in the United States with a rate of 14.7 per 1000 (1 in 68) for children aged 8 years old (CDC 2014). Individuals with autism often display a number of comorbidities including seizure disorder and intellectual impairment (Canitano 2007; Chakrabarti and Fombonne 2005). This chapter reviews the current evidence for dysfunction of the Reelin, GABAergic, and the FMRP -mGluR5 signaling pathways in autism. During development, the Reelin signaling system is instrumental in proper neuronal migration and brain lamination (Frotscher 1998; D’Arcangelo et al. 1995). In adults, Reelin is expressed in GABAergic interneurons (Pesold et al. 1998) and is involved in synapse formation and plasticity (reviewed by Stranahan et al. 2013). GABA is the most common inhibitory neurotransmitter in the brain. Disruption of the excitatory/inhibitory balance in the brain resulting from perturbations to this system may lead to the presence of seizures, abnormal information processing, and cognitive dysfunction associated with autism. Fragile X syndrome (FXS) is one of the most common forms of mental retardation and is often comorbid with autism. Behavioral deficits such as social anxiety, gaze avoideance, hyperarousal to sensory stimuli are common to both autism and FXS. In this chapter, we review current data from gene association and postmortem studies which implicate these three signaling systems in the pathology of autism.

1 Reelin and Autism

Reelin is a secreted extracellular matrix glycoprotein (DeBergeyck et al. 1998) with multiple functions during brain development including mediating proper brain lamination (Boyle et al. 2011; Hamburgh 1963; Tissir and Goffinet 2003) and facilitation of neuronal cell migration (D’Arcangelo et al. 1995; Hadj-Sahraoui et al. 1996). In adults, Reelin signaling is involved in synapse formation and modulation of synaptic transmission (Beffert et al. 2005; Chen et al. 2005; Groc et al. 2007; Herz and Chen 2006; Qiu and Weeber 2007; Ventruti et al. 2011; Weeber et al. 2002). Full length Reelin has a molecular weight of 410 kDa, while several cleavage products including Reelin 330 kDa, Reelin 180 kDa, and other fragments including a 220 kDa as well as 100 kDa C-terminal entity can be identified using SDS-PAGE (Jossin 2008; Fatemi et al. 2005a,b; Ignatova et al. 2004; Lambert de Rouvroit et al. 1999; Smalheiser et al. 2000). Reeler mice, which lack Reelin expression, display deficits in long term potentiation (Marrone et al. 2006) as do heterozygous Reeler mice (HRM) which are characterized by an approximately 50 % reduction in normal levels of Reelin (Tueting et al. 2006; Qiu and Weeber, 2007). These mouse strains also display behavioral deficits relevant to autism including impaired executive function, increased anxiety and motor impulsivity, impaired fear-conditioned learning, and deficits in sensorimotor gating behavior as measured by prepulse inhibition (PPI) (Ammassari-Teule et al. 2009; Barr et al. 2008; Ognibene et al. 2007; Qiu and Weeber, 2007; Tueting et al. 1999). It should be noted, however, that other groups have found no difference in PPI between HRM and wild type mice (Barr et al. 2008; Podhorna and Didriksen 2004; Teixeira et al. 2011) .

Reelin binds to three receptors: very low density lipoprotein receptor (VLDLR), apolipoprotein E receptor 2 (APOER2) (D’Arcangelo et al. 1999) and α3β1 integrin . Experiments involving Vldlr and Apoer2 knockout (KO) mice have demonstrated differing roles for each receptor with regard to neuronal migration with APOER2 enabling migration while Reelin binding to VLDLR may cease neuronal migration (Hack et al. 2007). Associated with APOER2 and VLDLR are ephrin B proteins (EFNBs) (Sentürk et al. 2011; Bouché et al. 2013). Reelin binding results in clustering of APOER2, VLDLR, and EFNBs and activation of FYN tyrosine kinase (Hiesberger et al. 1999; Strasser et al. 2004) and promoting the phosphorylation of disabled 1 (DAB1), a cytoplasmic adaptor protein (Hiesberger et al. 1999; Sentürk et al. 2011). Loss of function of EFNBs results in reduced phosphorylation of DAB1 and impaired neuronal migration (Sentürk et al. 2011). Phosphorylation of DAB1 activates a kinase cascade including phosphatidylinositol-3-kinase (PI3K) and protein kinase B (PKB/AKT) ultimately leading to the phosphorylation and inhibition of glycogen synthase kinase 3-beta (GSK3β) (Beffert et al. 2002; Hiesberger et al. 1999). Mouse embryos that express a nonphosphorylated Dab1 mutation display deficits in migration of sympathetic preganglionic neurons that is similar to what is observed in the Reeler mouse (Yip et al. 2007a). DAB1 phosphorylation also leads to the recruitment of the lissencephaly 1 (LIS1) complex which is important in both neuronal migration and proper cortical lamination (Assadi et al. 2003). GSK3β phosphorylates tau, a microtubule stabilizing protein (Hiesberger et al. 1999; Kwok et al. 2005). Inhibition of GSK3β leads to reduced phosphorylation of tau and allows for altered microtubule dynamics promoting neuronal cell migration. Figure 16.1 summarizes the Reelin signaling system .

Fig. 16.1
figure 1

The Reelin signaling cascade. Reelin binds to its extracellular receptors APOER2, VLDLR and α3β1 integrin resulting in clustering of the receptors and ephrin B proteins (ENFBs). This clustering recruits DAB1 proteins and activation of FYN kinase which phosphorylates DAB1. Phosphorylation of DAB1 leads to: (1) recruitment and activation of a kinase cascade involving PI3K and PKB/AKT1 which ultimately inhibits GSK3β; and (2) recruitment of LIS1 which promotes proper brain lamination. Inhibition of GSK3β results in dephosphorylation of tau which in turn destabilizes microtubule dynamics promoting neuronal migration. The binding of Reelin to its receptors and subsequent activation of FYN leads to phosphorylation of NMDA receptor subunits NR2A and NR2B. As a result, there is an inflow of calcium which induces long term potentiation (LTP) and synaptic plasticity. (Reprinted from Folsom and Fatemi 2013 Copyright (2013) with permission from Elsevier)

Reelin contributes to synapse formation and modulation of synaptic transmission via regulation of Ca2+ entry through N-methyl-d-aspartate (NMDA) receptors (Fig. 16.1) (Beffert et al. 2005; Chen et al. 2005; Groc et al. 2007; Herz and Chen 2006; Qiu and Weeber 2007; Ventruti et al. 2011; Weeber et al. 2002). Synaptic transmission in rodent models is impaired by mutations involving Apoer2 (Beffert et al. 2005, 2006) or loss of Dab1 (Trotter et al. 2013). Moreover, loss or transient downregulation of Dab1 results in impairments of associative learning, memory deficits, and impaired sensory motor gating as measured by prepulse inhibition, all of which are relevant to the pathology of autism (Teixeira et al. 2014; Trotter et al. 2013) .

Based on its role in brain development and synaptic transmission, the gene that codes Reelin (RELN) has been investigated as an autism candidate gene. An initial discovery found that an increase in the number of GGC triplet repeats in the 5′ untranslated region immediately before the RELN initiation codon, conferred vulnerability to the development of autism (Persico et al. 2001). RELN alleles with at least 11 triplet repeats in this region were preferentially transmitted to subjects with autism (Lugli et al. 2003; Persico et al. 2001). Subsequent studies in various population samples have verified an association between this region and autism (Dutta et al. 2007; Kelemenova et al. 2010; Skaar et al. 2005; Zhang et al. 2002). However, several other studies found no such association (Bonora et al. 2003; Devlin et al. 2004; Krebs et al. 2002; Li et al. 2004) .

A number of variants and single nucleotide polymorphisms (SNPs) of RELN have also been associated with autism in various populations including g.504742G > A in a Han Chinese sample (Tian 2012); rs2073559 in a Caucasian population sample (Ashley-Koch et al. 2007); rs736707 from intron 59 and rs362691 from exon 22 in a Caucasian population sample (Serajee et al. 2006); rs736707 in a Han Chinese population sample (Li et al. 2008), a finding that was recently confirmed in a South African population sample (Sharma et al. 2013). However, other groups have found no association between SNPs of RELN, including the ones listed above, and autism. A study of six RELN SNPs (rs727531, rs2072403, rs2072402, rs362691, rs362719, rs736707) in an Indian population sample failed to find a significant association for any of the SNPs and autism (Dutta et al. 2008). In contrast to the finding by Li et al. (2008), a separate study failed to find an association between rs736707 and autism in a Han Chinese population sample and, moreover, found no association for rs2229864, rs362691, and rs2073559 and autism (He et al. 2011) .

Reelin levels have been shown to be reduced in sera and in brains of subjects with autism (Fatemi et al. 2001, 2002, 2005a; Keller et al. 2000; Lugli et al. 2003). In sera from subjects with autism, reductions of Reelin 410 kDa have been observed (Fatemi et al. 2002; Keller et al. 2000). Lugli et al (2003) found that Reelin 330 kDa was significantly reduced by 25 % in sera of subjects with 11 or more CGG triplet repeats, suggesting that transmission of long alleles resulted in reduced Reelin expression. In brain, Fatemi et al (2001) found reduced expression of Reelin 180 kDa in cerebellum of subjects with autism when compared with healthy controls. A follow up experiment found reduced expression of Reelin 410 kDa, Reelin 330 kDa, and Reelin 180 kDa in prefrontal cortex of subjects with autism and reduction of Reelin 410 kDa and Reelin 180 kDa in cerebellum of subjects with autism, while there were no significant differences in parietal cortex of subjects with autism (Fig. 16.2) (Fatemi et al. 2005a). RELN mRNA was similarly reduced in both BA9 and cerebellum of subjects with autism as was mRNA for DAB1, while mRNA for VLDLR was significantly upregulated in both areas (Fatemi et al. 2005a). In addition to autism, Reelin downregulation has been observed in subjects with schizophrenia and mood disorders (Eastwood and Harrison 2003, 2006; Fatemi et al. 2000, 2005b; Guidotti et al. 2000; Impagnatiello et al. 1998) .

Fig. 16.2
figure 2

The gel mobility of Reelin bands of 410, 330, and 180 kDa in BA9 (a), cerebellar (b), and BA40 (c) homogenates of representative control (left histogram bar) and autistic adults (right histogram bar). a In BA9, there were significant reductions of Reelin 180 kDa/β-actin (p< 0.005), Reelin 330 kDa/β-actin (p < 0.035) and Reelin 410 kDa/β-actin (p < 0.035) versus control subjects. b In cerebella of subjects with autism, significant reductions in expression of Reelin 180 kDa/β-actin (p < 0.008) and Reelin 410 kDa/β-actin (p < 0.016) were observed versus control subjects. c In BA40, non-significant reductions were observed for full length reelin and the 330 kDa and 180 kDa fragments were observed. (Reprinted in a modified form Fatemi et al. 2005a. Copyright (2005) with permission from Elsevier)

2 GABA and Autism

Gamma-aminobutyric acid (GABA) is the main inhibitory neurotransmitter in the brain. GABA has many important roles in brain development including stimulating the proliferation of neural progenitor cells, migration and differentiation of neurites, and synapse formation (reviewed by Pizzarelli and Cherubini 2011). Approximately 20 % of all central nervous system (CNS) neurons are GABAergic (Somogyi et al. 1998).

GABA is synthesized from glutamate by the enzyme glutamic acid decarboxylase. GABA is then transported to synaptic vesicles by the GABA transporter vGAT. When an action potential arrives at the nerve terminal, GABA is released into the synaptic cleft where it binds to its postsynaptic receptors. GABA receptors are either ionotropic (GABAA) which are ligand-gated ion channels or metabotropic (GABAB) which are G-protein coupled receptors. GABAA receptors are responsible for mediating the fast inhibitory action of GABA (Brandon et al. 2000). GABAA receptors are also the sites for clinical action of a number of drugs including benzodiazepines, barbiturates, and anesthetics. Thus far, 19 GABAA receptor subunits have been characterized (α1–α6, β1–β3, γ1– γ3, δ, ε, π, θ, and ρ1–ρ3) (Ma et al. 2005; Brandon et al. 2000) which combine to form the heteropentameric GABAA receptors. The most common GABAA receptors consist of two α subunits, two β subunits, and one γ subunit. GABAB receptors are heterodimeric, formed from two subunits: GABAB receptor 1 (GABBR1) and GABAB receptor 2 (GABBR2) (Jones et al. 1998). GABAB receptors require one GABBR1 and one GABBR2 subunits in order to be functional. GABAB receptors contribute to synaptic events in the mammalian brain presynaptically by facilitating the release of neurotransmitters, including dopamine, serotonin, glutamate and GABA (Steiniger and Kretschmer 2003; Takahashi et al. 2010; Waldmeier et al. 2008). Postsynaptically, stimulation of GABAB receptors results in the generation of slow, long-lasting, inhibitory potentials (Bowery 2000; Kuriyama et al. 2000).

Recent studies have investigated levels of GABA in brain and plasma of subjects with autism (Gaetz et al. 2013; Rojas et al. 2013; Russo 2013). Plasma levels of GABA have been shown to be significantly increased in children with autism and that high levels of GABA correlated significantly with increased hyperactivity and impulsivity, tip toeing severity, light sensitivity, and tactile sensitivity (Russo 2013). Rojas et al (2013) found that ratios of GABA to creatine (Cr) were significantly reduced when compared to controls in the perisylvian region of the left hemisphere as visualized by single-voxel, point resolved spectroscopy. Interestingly, unaffected siblings also displayed reduced GABA/Cr ratios when compared with controls (Rojas et al. 2013). A second study, using the same procedure, found reduced GABA/Cr ratios in motor cortex and auditory cortex regions of interest (ROI) but not in the visual cortex ROI (Gaetz et al. 2013). The reduction of GABA levels in the brains of subjects with autism supports the hypothesis that the excitatory/inhibitory balance is disrupted in this population, which might help to explain the presence of seizure disorders (Tuchman and Rapin 2002) as well as cognitive and behavioral deficits associated with autism.

Glutamic acid decarboxylase 65 and 67 kDa proteins (GAD65/67) are the rate limiting enzymes responsible for the conversion of glutamate to GABA. GAD65 is significantly downregulated in cerebellum of subjects with autism and GAD67 is significantly reduced in parietal cortex of subjects with autism (Fatemi et al. 2002). More recently, significant reductions in GAD65 mRNA in the cerebellar dentate nuclei and significant reductions in GAD67 mRNA in Purkinje cells in cerebella from subjects with autism have been observed (Yip et al. 2007b 2009). Reductions in GAD65/67 could result in excessive glutamatergic (excitatory) and reduced GABAergic (inhibitory) signaling in important brain circuits that connect the cerebellum with the frontal cortex. The major deficiencies in levels of GAD 65 and 67 kDa proteins in two important brain areas in autism may subserve deficiency in the availability of GABA , thus affecting important biological functions including learning and motor activity. Additionally, decreased levels of GAD 65 and 67 kDa proteins could impact negatively on normal processing of visual, somatic, motor, and memory information processing, and could also explain the observations of increased blood, platelet, brain, and CSF glutamate levels in autistic patients (Moreno-Fuenmayor et al. 1996; Moreno et al. 1992; Shimmua et al. 2011; Shinohe et al. 2006).

A number of studies have now demonstrated that GABA receptors are reduced in brains of subjects with autism (Blatt et al. 2001; Samaco et al. 2005; Guptill et al. 2007; Fatemi et al. 2009a, b, 2010a, 2014; Oblak et al. 2010, 2011). Blatt et al. (2001) demonstrated a significant decrease in GABAA receptor binding sites (3H-muscimol-labeled binding sites) and benzodiazepine receptor binding sites (3H-flunitrazepam-labeled binding sites) in hippocampus of subjects with autism. Guptill et al. (2007), expanded on these experiments to demonstrate that the decrease in 3H-flunitrazepam-labeled benzodiazepine binding sites was due to a decrease in binding site number (Bmax) rather than altered affinity to ligand binding (Kd). Reduced 3H-muscimol-labeled GABAA receptor binding sites have also been observed in the posterior cingulate cortex and fusiform gyrus (Oblak et al. 2011). Finally, 3H-CGP54626-labeled GABAB receptor binding sites have also been observed to be reduced in the anterior and posterior cingulate cortex and fusiform gyrus of subjects with autism (Oblak et al. 2010).

Consistent with the results of binding assays, GABAA and GABAB receptor subunit proteins have been shown to be reduced in brains of subjects with autism (Fatemi et al. 2009a, b, 2010a, 2014). A comprehensive series of experiments have examined protein expression for GABA receptor subunits (16 GABAA and two GABAB) in superior frontal cortex [Brodmann Area 9 (BA9)], parietal cortex (BA40), and cerebellum of subjects with autism vs. matched controls. In BA9, there were reductions in GABAA receptor alpha 1 (GABRα1), GABAA receptor alpha 4 (GABRα4), GABAA receptor alpha 5 (GABRα5), GABAA receptor alpha 6 (GABRα6), GABAA receptor beta 1 (GABRβ1), GABAA receptor beta 2 (GABRβ2), GABAA receptor delta (GABRδ), GABAA receptor epsilon (GABRε), GABAA receptor gamma 2 (GABRγ2), GABAA receptor rho 2 (GABRρ2), and GABAB receptor 1 (GABBR1) proteins in brains of subjects with autism (Fig. 16.3) (Fatemi et al. 2009a, b, 2010a, 2014). In BA40, we observed significant reductions in GABRα1, GABAA receptor alpha 2 (GABRα2), GABAA receptor alpha 3 (GABRα3), GABRα5, GABAA receptor beta 3 (GABRβ3), and GABBR1 proteins in brains of subjects with autism (Table 16.1) (Fig. 16.3) (Fatemi et al. 2009a, b, 2010a, 2014). GABRα1, GABRβ3, GABBR1, and GABAB receptor 2 (GABBR2) proteins were significantly decreased in cerebella obtained from subjects with autism vs. matched controls (Fig. 16.3) (Fatemi et al. 2009a, b) (Table 16.1) (Fig. 16.3). Consistent with these findings, a previous report has also demonstrated reduction in GABRβ3 in cerebella of subjects with autism when compared with controls (Samaco et al. 2005). More GABA receptor subunits were reduced in BA9 than in BA40 or cerebellum. Gene expression changes have previously been demonstrated to be more robust in cerebral cortex of subjects with autism than in cerebellum, which is consistent with our findings (Voineagu et al. 2011). Altered protein expression for GABAA and GABAB receptor subunits have also been observed in the brains of subjects with schizophrenia and mood disorders (Fatemi et al. 2011a, 2013a, b, 2014).

Fig. 16.3
figure 3

Representative samples of GABRα1 (51 kDa), GABRα2 (51 kDa), GABRα3 (55 kDa), GABRα4 (64 kDa), GABRα5 (52 kDa), GABRα6 (50 kDa), GABRβ1 (55 kDa), GABRβ2 (56 kDa and 52 kDa), GABRβ3 (56 kDa), GABRδ (51 kDa), GABRε (45 kDa), GABRγ2 (45 kDa), GABRγ3 (51 kDa), GABRπ (50 kDa), GABRθ (70 kDa), GABRρ2 (54 kDa), GABBR1 (108 kDa), GABBR2 (105 kDa), and β-Actin (42 kDa) in BA9, BA40, and cerebellum of subjects with autism (A) and matched controls (C). [Reprinted in a modified from Fatemi et al. 2009a Copyright (2009) with permission from Springer Science+Business Media; from Fatemi et al. 2011b. Copyright (2011) with permission from Springer Science+Business Media; and from Fatemi et al. (2010a). Copyright (2010) with permission from Springer Science+Business Media]

Table 16.1 Summary of mRNA and protein findings for selected GABAA and GABAB receptor subunits in BA9, cerebellum, and BA40 of subjects with autism vs. matched controls

mRNA species for the same GABAA and GABAB receptors were measured via qRT-PCR in BA9, BA40, and cerebellum of subjects with autism vs. controls using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin as housekeeping genes (Fatemi et al. 2010a, 2014). Increased mRNA was observed in BA9 of subjects with autism for GABRα6 while there were reduced mRNA species for GABRα2, GABRα3, GABRα4, GABRα5, GABRβ1, GABRβ3, GABRγ3, and GABAA receptor theta (GABRθ) (Table 16.1) (Fatemi et al. 2010a, 2014). In BA40, qRT-PCR revealed increased mRNA expression of GABRγ3 and GABBR1 and reduced mRNA expression for GABRα3 in subjects with autism (Table 16.1) (Fatemi et al. 2010a, 2014). qRT-PCR of cerebella from subjects with autism revealed significant upregulation of mRNA species for GABRα2, GABRα3, GABRα4, GABRα5, GABRβ1, GABRβ3, GABRγ2, GABRγ3, and GABRθ (Table 16.1) (Fatemi et al. 2010a, 2014). In contrast, there were reduced mRNA species for GABRα6, GABRβ2, and GABBR1 in cerebella of subjects with autism (Fatemi et al. 2010a, 2014). While mRNA expression changes were variable in each brain region, protein expression was consistently downregulated.

Gene association studies for GABA receptor subunit genes and autism have been equivocal. An analysis of 12 GABAA and two GABAB receptor subunit genes found that GABRA4 (which codes for GABRα4), potentially through interaction with GABRB1 (which codes for GABRβ1) was associated with autism (Ma et al. 2005), a finding that was subsequently confirmed (Collins et al. 2006). Several studies have implicated the 15q11.2–q13 region, which includes genes for GABRβ3, GABRα5, and GABRγ3 (GABRB3, GABRA5, GABRG3) with autism (Hogart et al. 2007; Maddox et al. 1999; McCauley et al. 2004). GABRB3 has been associated with autism in a Korean population sample (Kim et al. 2006). SNPs of GABRA5 and GABRB3 have been shown to be nominally associated with autism (McCauley et al. 2004) as have SNPs of GABRG3 (Menold et al. 2001). However, a study using a Japanese population sample failed to find significant associations between GABRB3, GABRA5, or GABRG3 and autism (Tochigi et al. 2007). Moreover, others have similarly found no association for these genes and autism (Ma et al. 2005; Kelemenova et al. 2010). Recently, truncating mutations of the genes that code for GABRα3 and GABRθ (GABRA3 and GABRQ) have shown an association with autism spectrum disorders (Piton et al. 2013). Taken together, these results suggest that association between GABA receptor genes and autism may be population specific.

Reduction in GABAA and GABAB receptor subunits may help explain comorbid seizure disorder and cognitive deficits present in subjects with autism. Seizure disorder was present in many of the medical histories of subjects with autism used in our postmortem studies (Fatemi et al. 2009a, b, 2010a, 2014). However, when analyzed as a confound, we did not find an impact of seizure disorder on our results (Fatemi et al. 2009a, b, 2010a, 2014). Animal models of seizure disorder have shown reduced expression of GABBR1 and GABBR2 in brain (Han et al. 2006; Princivalle et al. 2003; Straessle et al. 2003). GABBR1 KO mice display seizure disorder and memory deficits (Prosser et al. 2001; Schuler et al. 2001). Eplieptiform activity interferes with cognition by causing disturbances of vigilance, shifting attention, and language difficulties (Binnie 1993) phenomena that often occur in children with autism and epilepsy. It has been hypothesized that the regression of language skills in children between the ages of two and three with autism may be due to epileptiform activity (Canitano 2007).

3 FMRP and Autism

Autism and fragile X syndrome (FXS) share many commonalities including intellectual disability, presence of seizures, learning difficulties, social deficits, anxiety, decreased attention, poor eye contact, delayed language acquisition and disorders of expression. Previous reports have shown the presence of autistic behavior in 15–47 % of patients with FXS (Bailey et al. 1998; Hatton et al. 2006; Kau et al. 2004; Kauffman et al. 2004). Individuals with diagnoses of both autism and FXS display greater severity of symptoms (Kau et al. 2004; Philofsky et al. 2004) and greater impairment in nonverbal cognition and expressive language (Philofsky et al. 2004). Boys with autism and FXS also show more cognitive impairment, abnormal behavior, and less adaptive behavior when compared to those with FXS alone (Kau et al. 2004). Interestingly, autistic symptoms have been shown to improve with age in subjects with FXS (McDuffie et al. 2010). In addition to phenotypic overlap, recently identified biological substrates have been proposed that create intriguing avenues of scientific interest.

The fragile X mental retardation 1 (FMR1) gene is located to the X chromosome. Mutations in this gene, resulting in loss of function, are almost entirely responsible for the development of FXS. Under normal circumstances there are anywhere from 5-55 5’ CGG repeats in the untranlated portion of the gene (Fu et al. 1991). Individuals with 56–200 repeats are often found in FXS families but do not display clinical symptoms of FXS and are considered to have the FMR1 premutation (Bardoni et al. 2001). When more than 200 5′ CGG repeats are present, there is extensive methylation of the 5′ region, including the promoter of FMR1 resulting in transcriptional silencing of FMR1 and the presence of clinical symptoms of FXS (Pieretti et al. 1991). FMRP binds to approximately 5 % of all mRNAs expressed in brain (Darnell and Klann 2013) potentially controlling a large number of important processes. It has been hypothesized that reduction in FMRP expression leads to unregulated protein synthesis, induced by group 1 metabotropic glutamate receptors (mGluRs), which in turn is responsible for the multiple physical and cognitive pathologies of FXS (Bear et al. 2004; Dölen and Bear 2008).

Boys who have the FMR1 premutation, especially if they present as clinical probands, are more likely to have a comorbid diagnosis of autism than nonprobands or control siblings who lack the premutation (Chonchaiya et al. 2012; Farzin et al. 2006). Additionally, these probands displayed increased rates of seizure disorder (Chonchaiya et al. 2012) and attention deficit/hyperactivity disorder (Farzin et al. 2006). Carriers of the FMR1 premutation have also been shown to have reduced amygdala volume, reduced activation of the right amygdala during an emotion matching task, and higher ratings of autism spectrum symptoms (Hessl et al. 2011). The authors found that while reduced FMRP and increased FMR1 mRNA were associated with reduced activation, reduced FMRP expression was identified statistically as the primary factor associated with reduced amygdala activation (Hessl et al. 2011).

The FMR1 gene may be a candidate gene for autism. The Xq27-q28 region, which includes FMR1 and methyl CpG binding protein 2 (MECP2), has shown some association with increased risk of autism (Vincent et al. 2005). A rare point mutation of FMR1 (A to C substitution at nucleotide 879 in exon 9) may contribute to autism and mental retardation in Japanese patients (Shinahara et al. 2004). An intronic variant of FMR1 (IVS10+14C-T) showed no evidence of increasing the risk of autism (Vincent et al. 2004). There may also be structural differences in the FMR1 gene in subjects with autism vs. controls. An investigation of CGG repeat length and AGG interruption of the FMR1 gene in subjects with infantile autism vs. healthy controls found that in the case of infantile autism, there were less AGG interruptions (Poon et al. 1998). This pattern is similar to what is found in subjects with FXS, although none of the study subjects had a comorbid diagnosis of FXS (Poon et al. 1998).

A recent series of experiments involving postmortem brain tissue have shown dysregulation of FMRP , mGluR5, and downstream targets of FMRP-mGluR5 signaling in cerebellar vermis and PFC in subjects with autism (Fatemi and Folsom 2011; Fatemi et al. 2011b, 2013c; Rustan et al. 2013). Levels of FMRP were reduced in PFC and cerebellar vermis of adults with autism while there were no differences in these brain regions in children with autism (Fig. 16.4) (Fatemi and Folsom 2011; Fatemi et al. 2011b). Phosphorylated FMRP (p-FMRP) is reduced in cerebellar vermis of adults and children with autism and in PFC of adults with autism (Rustan et al. 2013). Dephosphorylation of FMRP leads to its ubiquitination and subsequent degradation, a process that may be the result of overactive mGluR5 signaling (Nalavadi et al. 2012). Indeed, in the same tissues, increased expression of mGluR5 was observed in PFC and cerebellar vermis of children with autism (Fig. 16.4) (Fatemi and Folsom 2011; Fatemi et al. 2011b). Interestingly, none of the subjects with autism had a comorbid diagnosis of FXS (Fatemi and Folsom 2011; Fatemi et al. 2011b, 2013c). Moreover, similar studies performed in subjects with schizophrenia and mood disorders found reduced expression of FMRP and mGluR5 in lateral cerebellum of subjects with schizophrenia, bipolar disorder, and major depression and in BA9 of subjects with schizophrenia and bipolar disorder (Fatemi et al. 2010b, 2013a, b). These results indicate that dysfunction in FMRP and mGluR5 expression may be common to multiple psychiatric disorders.

Fig. 16.4
figure 4

Expression of FMRP and mGluR5 in brains from subjects with autism vs. matched controls. a) Ratio of FMRP/NSE in cerebellar vermis. b) Ratio of mGluR5/NSE in cerebellar vermis. c) Ratio of FMRP/NSE in BA9. d) Ratio of mGluR5/NSE in BA9. Ratios are expressed as mean ± standard error *, p < 0.05. [Reprinted in a modified form Fatemi et al. 2011b Copyright (2011) with permission from John Wiley and Sons and from Molecular Autism; Fatemi and Folsom 2011. Copyright (2011) with permission from BioMed Central]

FMRP is a negative regulator of mGluR5. It has been proposed that in the absence of FMRP, mGluR5-dependent protein synthesis goes unchecked, resulting in the anatomical and physical deficits associated with FXS (Bear 2005; Bear et al. 2004; Dölen and Bear 2008). Evidence from FMR1 KO mice support this theory including: (1) increased long-term depression (LTD) (Huber et al. 2002); (2) an increased number of immature dendritc spines (Grossman et al. 2006); and (3) increased epileptiform activity (Yan et al. 2005), all of which are present in individuals with FXS (reviewed by Garber et al. 2008). Importantly, use of allosteric inhibitors of mGluR5 such as such as 2-methyl-6-(phenylethynyl)-pyridine (MPEP), or lowering of the concentration of mGluR5 reverse structural and behavioral deficits in FMR1 KO mice including the number of dendritic spines, deficits in prepulse inhibition, and presence of seizure that are also present in autism (de Vrij et al. 2008; Dölen et al. 2007; Westmark et al. 2009; Yan et al. 2005; Yuskaitis et al. 2010). The use of allosteric modulators of mGluR5 as well as other treatments for FXS and autism has been an ongoing line of research (Berry-Kravis et al. 2011; Hagerman et al. 2012; Li et al. 2013).

Four downstream targets of FMRP-mGluR5 signaling have also been investigated: homer 1, amyloid beta A4 precursor protein (APP), ras-related C3 botulinum toxin substrate 1 (RAC1), and striatal-enriched protein tyrosine phosphatase (STEP) (Fatemi et al. 2013c). These proteins are involved in synapse formation and neural plasticity (APP); regulation of N-methyl-D-aspartate (NMDA) receptor function (STEP); synaptogenesis, receptor trafficking, and involvement in dopaminergic and glutamatergic signaling (homer 1); and modulation of dendritic spine morphology and density (RAC1), all of which are relevant to autism (Goebel-Goody et al. 2012; Nakayama et al. 2000; Priller et al. 2006; Szumlinski et al. 2006; Turner et al. 2003). Protein levels of RAC1, APP 120 kDa and APP 88 kDa were upregulated in BA9 of children with autism (Fatemi et al. 2013c). In BA9 of adults with autism, there was increased protein expression of RAC1 and STEP 46 kDa while there was reduced expression of homer 1 (Fatemi et al. 2013c). In cerebellar vermis of adults with autism there was significantly increased RAC1 protein expression, while there was significantly reduced expression of APP 120 kDa, STEP 66 kDa, STEP 27 kDa, and homer 1 (Fatemi et al. 2013c). In contrast, there were no changes observed in cerebellar vermis of children with autism (Fatemi et al. 2013c).

The reduced expression of FMRP in subjects with autism may help explain reduced expression of GABA receptor subunits and Reelin. In animal models of FXS, expression of GABAA receptor subunits have been shown to be reduced or eliminated by the loss of function of the FRM1 gene and consequent loss of FMRP (D’Hulst et al. 2006; El Idrissi et al. 2005, Gantois et al. 2006). El Idrissi et al. (2005) found reduced expression of the GABAA β subunit in cortex, hippocampus, brainstem, and diencephalon of fragile X (FraX) mice. Gantois et al (2006) found reduced GABRδ mRNA in hippocampus and neocortex of Fmr1 knockout (KO) mice. A separate study found reduced mRNA for GABRα1, GABRα3, GABRα4, GABRβ1, GABRβ2, GABRγ1, and GABRγ2 in cortex, but not cerebellum of Fmr1 KO mice (D’Hulst et al. 2006).

Reelin has also been identified as a downstream target of FMRP (Darnell et al. 2011). Altered expression of FMRP in subjects with autism may impact levels of Reelin as well.

4 Conclusions

Autism is a heterogeneous disorder in which multiple signaling systems are impacted. The Reelin, GABAergic, and FMRP-mGluR5 signaling systems separately, and perhaps synergistically, contribute to the pathology of autism. Dysfunction of the Reelin signaling system may result in abnormalities in brain morphology associated with autism as well as dysfunctional synaptic transmission. GABAergic system dysfunction could result in cognitive impairments as well as seizure disorder. Deficits in the FMRP-mGluR5 signaling system contribute to intellectual impairment, altered neuronal structure and seizure disorder. Reelin is known to regulate GABAergic and glutamatergic neurotransmission (Tissir and Goffinet 2003; Marrone et al. 2006). FMRP regulates GABAA receptor expression (D’Hulst et al. 2006; El Idrissi et al. 2005, Gantois et al. 2006) and targets Reelin (Darnell et al. 2011), potentially regulating its expression as well. Dysfunction in one system may lead to dysfunction in other systems. Interplay between these systems may result in multiple abnormal phenotypes associated with autism. These systems also identify targets for therapeutic intervention which may ameliorate multiple symptoms of autism.