Introduction
Autism spectrum disorder (ASD) is a neurodevelopmental disability that manifests in early childhood and is characterized by deficits in social skills, behaviors, and communication. According to the World Health Organization, approximately 1 in 160 children worldwide [1] and about 1 in 44 children in the United States have ASD [2], which can occur in all racial and ethnic groups, and is four times more prevalent in boys than in girls [3]. Individuals with ASD can have co-occurring conditions, such as attention deficit hyperactivity disorder (ADHD), bipolar disorder, depression, intellectual disability, language and developmental delays, speech disorder, and gastrointestinal symptoms [4]. Although the cause of ASD is ambiguous, genetic and non-genetic factors most likely contribute to its development [5].
ASD is associated with several genetic syndromes, a high incidence of chromosomal rearrangements, and the presence of common and rare variants [6]. Methodologic advances have revealed that common, heritable polygenic risk accounts for ~50% of ASD cases; major-affect mutations account for 15%; and rare de novo copy number variations (CNVs) and single-nucleotide variants (SNVs) that alter the structural genome account for ~5% [7]. No theory posits a clear unifying mechanism of ASD at the molecular or cellular level, because it remains unclear whether ASD is many disorders converging on a few molecular pathways or a few disorders with complex, diverse mechanisms [8].
The cellular and molecular bases of autism can be attributed to increased local connectivity in brain regions, neuronal migration deficits, excitatory/inhibitory imbalance, and synaptic dysregulation [9,10,11,12]. Many studies have highlighted the genetic heterogeneity underlying ASD and indicated that several ASD-associated gene or protein products interact with neuronal, synaptic, and other neurodevelopmental pathways [13, 14]. Neurologic disorders, such as ASD, cause microdamage to the brain, and detection of the resulting structural and functional changes requires the use of high-resolution, noninvasive imaging techniques, such as magnetic resonance imaging (MRI). Furthermore, neuroimaging studies have provided evidence of altered cortical and subcortical structures, impaired white matter (WM) connectivity, and atypical connectivity in the frontal and temporal brain regions involved in various cognitive functions [15].
Because genes directly affect brain development and function, genetic polymorphisms or aberrations might be strongly associated with the functioning of the compromised neural systems and behavioral outcomes [16]. Neuroimaging can be used to investigate the effect of genetic variations on brain structure, function, and connectivity; this approach is known as “neuroimaging genetics” [17]. Neuroimaging genetics can delineate the molecular mechanisms induced by genetic variants (common and rare) linked to neurodevelopmental disorders (NDDs). Neuroimaging genetics enables us to investigate gene-specific effects on different functional brain systems, which will contribute to future diagnosis of various NDDs, including ASD.
In this review, we will explore various neuroimaging techniques that can be used to assess the impact of genetic factors on brain structure, function, and metabolism. In addition, we will discuss the neuroimaging genetics approach can be used to identify novel biomarkers for the early diagnosis of ASD.
Brain changes associated with genetic changes in ASD
Structural and functional changes
Several gene polymorphisms have been linked to structural and functional changes in the brains of individuals with ASD. For example, Homeobox (HOX) genes that play an important role in defining cell identity and positioning during embryonic development are also associated with autism [18]. An earlier study has shown aberrant hindbrain development and craniofacial defects in HOXA1/HOXB1-mutant mice as a result of rhombomere misspecification within the hindbrain [19]. Other studies have shown the association of HOXA1 A218G polymorphism with increased head circumference in autistic individuals [20, 21]. On the other hand, HOXB1 alleles have been found to affect stereotypic behaviors and influence head growth rates, but to a much lesser extent than HOXA1 A218G in autistic individuals [22]. Current research suggests that approximately 25% of individuals with constitutional PTEN mutations might meet the criteria for ASD [23]. Autistic individuals with germline PTEN mutations D252G (exon 7), H93R (exon 4), and F241S (exon 7) were found to have increased head circumferences than other autistic subjects [24]. In another study, the de novo missense PTEN mutation D326N (exon 8) was identified in an autistic patient with developmental delay, mental retardation, and extreme macrocephaly; the patient also showed prenatal and postnatal overgrowth [25].
A recent systematic review summarized brain structural MRI (sMRI) findings in monogenic disorders that are strongly associated with ASD [26]. The review included mutations in PTEN, SHANK3, SYNGAP1, CHD8, ARID1B, ADNP, POGZ, MED13L, SLC6A1, and ANKDR11, which are associated with different brain abnormalities, most prominently in the WM, GM, and ventricular regions [26]. Studies have also reported the association of contactin-associated protein 2 (CNTNAP2) polymorphisms with WM and GM abnormalities [27]. CNTNAP2 is a master gene that causes speech-language delay and is central to the manifestation of autism [28]. A recent MRI study involving 118 individuals with ASD and 122 typically developing (TD) controls showed the association of a CNTNAP2 variant (rs2538991) with WM volume of the right anterior cingulate gyrus in ASD individuals [29].
MET receptor tyrosine kinase and its ligand, hepatocyte growth factor help mediate neurodevelopmental events that are associated with brain structural pattern and circuitry and has a pleiotropic role in multiple organs’ ontogenesis [30]. Functional polymorphism in MET gene has been associated with increased risk for autism [31]. A study involving 75 individuals with ASD and 87 TD controls showed an ASD-risk variant in the met receptor tyrosine kinase gene to be associated with altered WM connectivity in individuals with ASD, relative to TD controls [32].
Mutations associated with the chromodomain helicase DNA-binding protein 8 (CHD8) gene are also implicated in autism. CHD8 is a transcriptional regulator that is involved in the remodeling of chromatin structure and is crucial for dendrite development and neuronal migration [33]. A case report study, using whole-exome sequencing (WES), identified a de novo mutation of the CHD8 gene in a clinical ASD phenotype including intellectual disability (ID), macrocephaly, and craniofacial abnormalities observed in a boy with developmental delay [34].
In a study comparing mouse models of autism to wild-type (WT) controls, NLGN3- and MECP2-mutant mice showed increased cerebellar volumes, and ITGB3-mutant mice showed reduced cerebellar volume [35]. Functional analysis of cortical neurons in MECP2-mutant mice showed abnormal growth of dendrites and axons, suggesting that MECP2 mutations impair neuronal development which might lead to ASD [36]. Mutations associated with the SH3 and multiple ankyrin repeat domains 3 (SHANK3), a synaptic scaffolding protein required for synaptic functioning, have been implicated in ASD, with knockout (KO) mice having reduced total brain volume, hippocampus, and thalami, and enlarged basal ganglia [37]. Loss of SHANK3 has been associated with altered prefrontal functional connectivity in mice, suggesting that this deletion impairs social and communication behaviors and contributes to ASD pathogenesis [38].
Neurexins are presynaptic cell-adhesion proteins that are involved in synapse formation. Single-cell RNA sequencing analysis on induced pluripotent stem cells-derived neural stem cells revealed that, compared to neurons of a healthy patient, those of an autistic patient carrying the biallelic neurexin 1-alpha (NRXN1-α) deletion had impaired maturation of action potentials and decreased calcium signaling [39]. Additionally, diffusion MRI analyses of the brain tissues of NRXN2-α–KO mice showed altered microstructures in the social brain regions and impaired structural connectivity between the amygdala and orbitofrontal cortex regions, suggesting a role of NRXN2 in altering social behaviors [40].
In another study, homozygous CNTNAP2–/– mice exhibited reduced long-range and local functional connectivity in the prefrontal and midline brain regions, suggesting that homozygous loss-of-function mutations in CNTNAP2 predispose individuals to NDDs, such as autism [41]. Another study using a forebrain organoid model generated from induced pluripotent stem cells of patients with syndromic ASD carrying the homozygous CNTNAP2 c.3709DelG mutation showed that the mutation causes cortical overgrowth in the organoids that was rescued by repairing the pathogenic mutation via CRISPR–Cas9, thus confirming the causative effect of homozygous CNTNAP2 mutation in ASD [42].
CD38, a transmembrane protein plays an important role in controlling social behaviors due to its role in regulating oxytocin secretion processes [43]. Two single-nucleotide polymorphisms (SNPs) in the CD38 gene (rs3796863 and rs1800561) have been detected in individuals with ASD [44]. Plasma levels of oxytocin were lower in individuals with ASD carrying the R140W allele than in those lacking the allele [44]. Treating a proband carrying the R140W allele with intranasal oxytocin improved social, communication, and emotional behaviors [44].
Along with CD38, oxytocin receptor (OXTR) genes influence social behavior, and OXTR mutations are a risk factor for ASD [45]. Genotyping for SNPs and fMRI analysis of 38 adolescents with high-functioning autism (HFA) and 33 TD controls [46] showed OXTR SNPs association with brain activation within the right supramarginal gyrus and inferior parietal lobule during an emotion-recognition task in autistic individuals [46]. Another study involving 209 probands with ASD investigated the influence of two polymorphisms (rs1042778, rs53576) in OXTR on ASD-related clinical symptoms including panic and aggressive behaviors [47]. The presence of OXTR rs1042778 T allele was associated with panic and aggressive behaviors in individuals with ASD, suggesting the importance of OXTR in ASD diagnosis and clinical phenotypes [47]. The disparity in the prevalence of ASD among males and females gives rise to a sex bias, a concept that is poorly understood in the neurobiology of autism. In relation to this, an imaging-genetics study consisting of 50 females with HFA and 52 females as TD controls and 37 males with HFA and 34 males as TD controls between the ages of 8 and 17 assessed the impact of ASD-associated OXTR variants on reward network functional connectivity in both male and female subjects with ASD [48]. Females carrying more ASD-associated OXTR variants showed increased connectivity between reward-related brain regions (nucleus accumbens) and the prefrontal cortex region, compared to males with ASD [48].
Genetic variants of arginine vasopressin receptor 1 A (AVPR1A) gene have also been linked with autism [49]. In a study involving 121 healthy volunteers, a functional imaging task was performed to assess the association between AVPR1A genetic variants and amygdala activation [49]. Carriers of the AVPR1A ASD-risk alleles (RS1 and RS3) had differential activation of the amygdala [49]. In another study involving 1104 healthy subjects, MRI, genotyping, and learning and memory assessment were performed to assess the effect of AVPR1A RS3-RS1 haplotypes on verbal learning and memory: individuals carrying the short alleles of RS3-RS1 haplotypes displayed poor verbal memory performance than those carrying the long alleles [50]. A study consisting of 212 ASD probands and their biological parents conducted a family-based association test to assess the effect of polymorphisms in the AVPR1A promoter region on social behavior [51]. The study found two AVPR1A SNPs (rs7294536 and rs10877969) to be over-transmitted as a risk allele in Korean families with ASD; suggesting their involvement in dysregulating social behavior and contributing to the pathophysiology of ASD [51].
Mutations associated with the reelin (RELN) gene, which encodes a large glycoprotein RELN that guides neuronal migration and positioning during embryonic development [52], have been implicated in ASD pathology [53, 54]. A de novo RELN R2290C mutation identified in an ASD proband in a conserved arginine-amino acid-arginine domain were found to impair RELN protein secretion and these effects were recapitulated in a heterozygous RELN mouse mutant model [55]. Analysis of RELN R2290C heterozygous neurospheres revealed an upregulation in Protein Disulfide Isomerase A1, a chaperone protein responsible for the formation of disulfide bonds [55]. In contrast, a study comparing plasma RELN levels between 40 ASD and 19 healthy children found higher RELN levels in children with ASD than in healthy controls [56].
The gene T-brain 1 (TBR1) encodes the transcription factor TBR1, which mediates gene transcription and cortical neurogenesis and is crucial for normal neurodevelopment. Several preclinical and clinical studies have shown TBR1 to be implicated in ASD [57,58,59,60,61]. TBR1 haploinsufficiency results in defective axonal projections of amygdala neurons and impairs social interaction, memory, cognitive flexibility [58], and neuronal activation of the olfactory system in mice models [57]. Additionally, mice carrying the heterozygous TBR1 K228E mutation showed altered cortical development, increased levels of TBR1, inhibitory synaptic transmission, and ASD-like behavioral phenotypes [60]. De novo and missense TBR1 mutations also disrupt TBR1 functions, such as subcellular localization and transcriptional repression, in individuals with sporadic ASD [61]. Moreover, TBR1 interacts with FOXP2, which is associated with speech and language disorders and the TBR1–FOXP2 interaction was shown to be abolished in patients with sporadic ASD [61].
Ankyrin 2 (ANK2), an important gene that encodes for ankyrin B (ankB) protein is involved in membrane stabilization and localization of ion channels and transporters. ANK2 mutation in mice has been found to increase axon branching and ectopic connectivity and impairs social and communication behaviors [62]. ANK2 is highly expressed during early neurodevelopmental stages and is an important regulator of neurogenesis [63]. Loss of ANK2 impaired differentiation of neural stem cells to neurons and altered the expression of genes involved in neural development, suggesting ANK2 haploinsufficiency as a risk factor for ASD [63].
Mice carrying the NLGN3 R351C mutation showed delayed synapse elimination in the cerebellum suggesting NLGN3 involvement in synapse refinement of the cerebellar circuitry that might be associated with ASD pathogenesis [64]. Additionally, NLGN3 R451C–mutant mice displayed more aggressive and repetitive behaviors than WT controls [65]. Another study observed that NLGN3-knockin mice exhibited reduced GM and WM volumes and reduced social and anxiety-related behaviors compared to WT controls [66].
These studies provide evidence that monogenic risk factors for autism share common involvement of the prefrontal cortex, cerebellum, amygdala, and hippocampus. In addition, most of the genetic mutations responsible for structural and functional brain changes converge on biological pathways that are involved in corticogenesis, synaptogenesis, chromatin modification, and transcriptional and translational processes, all of which contribute to social and behavioral impairments, the core deficits associated with ASD.
Metabolic changes
Proper regulation of cellular metabolism is essential for maintaining cellular function, which is crucial for the central nervous system (CNS), specifically for the brain, where energy consumption and metabolic changes are dynamic [67], and metabolic changes in neurons are critical for neuroplasticity and cognitive functions [67]. The heterogeneous and multifaceted pathological nature of ASD clearly explains why the genes affecting brain metabolism have been under-investigated. Nevertheless, to refine the search for metabolic biomarkers in ASD, the effect of candidate genes involved in ASD must be explored in metabolic pathways specifically affecting brain energy metabolism and neuron-astrocyte interactions to provide insight into which metabolic pathways are disrupted in ASD and how to target those pathways via neuroimaging genetics.
Autism is a multifactorial disease, with many candidate genes in its etiology. However, only a few of those genes such as such as DISC1, SHANK3, ITGB3, SLC6A4, RELN, RPL10, and AVPR1α are found to be associated with brain metabolism [68]. Moreover, several metabolic pathways are found to be altered in the prefrontal cortex of ASD individuals [69]. These metabolic pathways include glutathione metabolism, galactose metabolism, purine and pyruvate metabolism, starch and sucrose metabolism, arginine and proline metabolism, cysteine and methionine metabolism, propanoate metabolism, nicotinate, and nicotinamide metabolism, and the tricarboxylic acid (TCA) cycle [69].
The RNA-binding fox 1 (RBFOX1) gene is associated with various neuropsychiatric disorders (e.g., ASD, ADHD, epilepsy, intellectual disability, and schizophrenia) [70,71,72,73]. A study has also highlighted the role of cytoplasmic RBFOX1 in regulating the expression of genes involved in synaptic transmission and autism [74]. No association of RBFOX1 with brain metabolism in ASD has been found, but its role in Alzheimer’s disease suggests its exploration in ASD.
Mitochondrial dysfunction is one of the most common metabolic abnormalities in ASD [75], and lactate and pyruvate are important biomarkers for mitochondrial energy metabolism [76]. Many studies have shown that lactate dehydrogenase A (LDHA) and B (LDHB) are involved in ASD pathophysiology [77,78,79,80,81].
Glucose fuels neuronal oxidative metabolism by providing ATP and the precursors required for neurotransmitter synthesis; glucose is also transported across the blood–brain barrier and into neurons by facilitative glucose transporters [82]. Glucose transporter-1 (GLUT-1) is predominantly found in the endothelial cells of the blood–brain barrier; GLUT-3 is the main transporter expressed in neurons [83]. Increased mRNA levels of GLUT-1, GLUT-3, and three key enzymes in glucose metabolism (hexokinase 1, pyruvate kinase, and pyruvate dehydrogenase) have been observed in the contralateral brain area of a mouse model of traumatic brain injury (TBI) [84]. Additionally, increased expression of lactate transporter in astrocytes and reduced expression of neuronal MCT-2 has been observed in the ipsilateral cortex and hippocampus of the TBI mouse model, suggesting that sustained impairment of glucose metabolism after TBI is neuron-specific [84]. Neuronal GLUT-3-deficient heterozygous mice demonstrated ASD-like features, and this phenotype was associated with increased GLUT-1 and MCT-2 concentrations suggesting that the neuronal glucose deficiency was compensated for enhanced uptake of lactate by the brain [85]. Despite this metabolic compensation, neuronal function was altered in GLUT-3–heterozygous mice, as observed by increased electroencephalographic seizure activity, neurobehavioral abnormalities (i.e., abnormal spatial learning and working memory), and deficits in social behavior, all features observed in ASD [85, 86]
In the adult CNS, cholesterol is derived through de novo synthesis by astrocytes [87]. A major constituent of cholesterol catabolism is the neuronal enzyme cytochrome P450 family 46 subfamily A member 1 (CYP46A1), which protects neurons and helps convert cholesterol to 24-hydroxycholesterol (24 HC), enabling it to cross the blood-brain barrier [88]. An indirect association of CYP46A1 has been shown in autistic children with high plasma levels of 24 HC [89]. Moreover, 24 HC plasma levels have been inversely correlated with age in autistic individuals [89].
The disrupted in Schizophrenia 1 (DISC1) gene is involved in neurodevelopmental processes, such as neuronal proliferation, differentiation, and migration [90]. Prenatal disruption of DISC1 in fetal neural progenitor cells of the dominant-negative DISC1 (DN-DISC1) adult mice has been found to cause significant anxiety and depression-like behavioral changes [91], elevated levels of GABA, and increased cell density of parvalbumin+ interneurons in the cingulate cortex, motor cortex, and the retrosplenial granular cortex [91]. Alternatively, somatostatin+ and neuropeptide-Y+ interneurons were found to be decreased in other brain regions, suggesting that disrupting DISC1 function affects the localization of interneuron subtypes [91]. Moreover, DN-DISC1 was found to interact with Dlx2 and negatively regulate Dlx2-mediated Wnt-signaling pathway activation [91]. Knocking down DISC1 and the expression of DN-DISC1 was found to decrease GLUT-4 mRNA and protein levels and reduce glucose uptake by primary astrocytes, which was associated with reduced oxidative phosphorylation, glycolysis, and lactate production in vitro and in vivo [92]. Moreover, treatment with lactate rescued the behavioral abnormalities in DN-DISC1 mice [92]. Thus, altered DISC1 function in astrocytes contributes to metabolic abnormalities that might cause cognitive and behavioral deficits in various neuropsychiatric disorders [92]. In a recent study, deletion of AUTS2 was found to impair social interactions, reduce uptake of brain glucose, and inhibit the pentose phosphate pathway in a conditional knockout mouse model with AUTS2 deletion (AUTS2-cKO) [86].
Obesity is a common risk factor associated with NDDs, such as ADHD and ASD [93, 94]. The adolescent medial prefrontal cortex region is found to be vulnerable to high-fat diets (HFDs) via RELN deficiency [95]. RELN acts through ApoER2 and very-low-density lipoprotein receptor (VLDLR) and plays an important role in cholesterol and fatty acid metabolism [96]. Hypothalamic levels of RELN protein and APoER2 and VLDLR mRNA were found to be altered in mice fed with HFD [97]. Moreover, the recombinant central fragment of RELN affects membrane potential and action potential firing by altering pre- and postsynaptic inputs on the arcuate nucleus satiety-promoting proopiomelanocortin (ARH-POMC) neurons in a POMC-EGFP mouse model, thus suggesting the role of RELN in mediating energy homeostasis by acting on ARH-POMC neurons [97].
The role of SHANK3 in synaptic function and development and its recognition as an important candidate gene in ASD has been well established [98], but that in cerebral metabolism has not been well studied. Increased rates of cerebral synthesis have been reported in several brain regions of the SHANK3-KO mice, thus indicating high protein turnover [99]. Moreover, increased pERK in hippocampal tissues, and reduced pERK/ERK and pmTOR/mTOR ratios in synaptosomal-enriched frontal cortex lysates suggested a loss of protein-synthesis regulation via these pathways [99].
The heterogeneous and multifaceted pathological nature of ASDs clearly explains why the genes affecting brain metabolism are under-investigated. To refine the search for metabolic biomarkers in ASD, the effect of candidate genes involved in ASD must be explored in different metabolic pathways specifically affecting brain energy metabolism and neuron-astrocyte interactions. Identification of these metabolic alterations in the brain can provide an insight into the metabolic pathways disrupted in ASD and help target those pathways using the “neuroimaging genetics” approach.
Role of imaging in detecting brain changes
Neuroimaging is vital to understanding the structure and functioning of the brain. In the field of ASD, most neuroimaging genetics studies are focused on structural and functional brain assessments. In this review article, we emphasize the imaging techniques used for monitoring changes in brain metabolism and neurotransmitter levels in individuals with ASD.
FOR MORE INFORMATION PLEASE GO TO LINK: