Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

This chapter is written to make the fast-paced, expanding field of the genetics of autism accessible to those practitioners who help children with autism. New genetic knowledge and technology have quickly developed over the past 30 years, particularly within the past decade, and have made many optimistic about our ability to explain autism. Among these advances include the sequencing of the human genome (Lander et al., 2001) and the identification of common genetic variants via the HapMap project (International HapMap Consortium, 2005), and the development of cost-efficient genotyping and analysis technologies (Losh, Sullivan, Trembath, & Piven, 2008). Improvement in technology has led to improved visualization of chromosomal abnormality down to the molecular level. The four most common syndromes associated with autism include fragile X syndrome, tuberous sclerosis, 15q duplications, and untreated phenylketonuria (PKU; Costa e Silva, 2008). FXS and 15q duplications are discussed within the context of cytogenetics. TSC is illustrated within the description of linkage analysis.

As recently reviewed (Li & Andersson, 2009), the first inkling of chromosomal anomaly causing clinical pathology occurred in 1959 when an extra copy of chromosome 21 was associated with the Down syndrome phenotype. Shortly thereafter, chromosomal abnormality was recognized for Patau’s syndrome (trisomy 13) and Edward’s syndrome (trisomy 18). The chromosomal banding technique developed in 1970 led to the identification of many structural chromosomal abnormalities associated with clinical conditions. The discovery that genes located close to each other on a chromosome are often inherited together led to linkage analysis in the 1970s and 1980s. Linkage techniques related chromosomal abnormalities with known genetic anomalies using an affected sibling–pair design in multiplex (having more than one affected member) families. This led to the identification of chromosomal abnormalities in such conditions as neurofibromatosis, tuberous sclerosis, and dyslexia (Smith, 2007). Improving technology in the 1990s enabled the detection of small genomic alterations of 50–100 kb and the direct visualization of these alterations in uncultured cells via fluorescent in situ hybridization (FISH). This technique ushered in the field of molecular genetics (Li & Andersson, 2009) and allowed the identification of chromosomal microdeletions and duplications in areas of the chromosome where there is already high suspicion that abnormality would exist. FISH enables prenatal and cancer genetics screening and has led to the identification of genetic aberrations associated with Angelman’s and Prader Willi syndromes.

In the past decade, microarray cytogenetics has permitted the study of the entire genome on a single chip with resolution as fine as a few hundred base pairs (Li & Andersson, 2009). Such microarray technology represents a union between molecular genetics and classical cytogenetics. Two types of microarray technology are used clinically: comparative genomic hybridization (CGH) and single-nucleotide polymorphism (SNP) analysis. CGH directly measures copy number differences between a patient’s DNA and a normal reference DNA spanning known genes, chromosomal regions, or across the entire genome. SNP analysis, on the other hand, provides identification of a single point mutation via sequencing the gene in areas of suspected abnormality, previously identified via CGH or FISH. Another offshoot of microarray technology is submicroscopic chromosome copy number variation (CNV) analysis, in which deletions or duplications involving > 1-kb DNA have been detected in patients with mental retardation, autism, and multiple congenital anomalies.

There are several recent, detailed reviews of the genetics of autism (Abrahams & Geschwind, 2008; Li & Andersson, 2009; Losh et al., 2008; O’Roak & State, 2008) and this chapter summarizes those reviews. The reader is encouraged to first review the overview of gene expression (Box 6.1) and essential nomenclature used in genetics (Box 6.2). In addition, one must understand basic concepts pertinent to brain development (Box 6.3) and to the neurobiology of autism (Box 6.4) in order to understand its genetics. For example, there is a compelling rationale that genetically directed mechanisms that regulate the assembly of the brain during embryogenesis, when gone awry, may cause autism (Costa e Silva, 2008).

Indirect Evidence of Heritability

Three lines of research indirectly attest to the heritability of autism: twin studies, family studies, and the fact that autism affects more boys than girls. In general, hereditability appears to be greater when a broader definition of autism (including individuals with cognitive deficits and/or social impairment) instead of the specific DSM-IV criteria for autism is used.

Twin Studies

In the first twin study of autism (Folstein & Rutter, 1977), the concordance rate in monozygotic (MZ) pairs (36%) was significantly greater than that found in dizygotic (DZ) pairs (0%). If the phenotype was expanded to include a cognitive or a language disorder, the concordance rates were 82 and 10%, respectively. Two subsequent studies found an MZ/DZ concordance rate of 91–0% (Steffenburg et al., 1989) or 60–5% utilizing the specific phenotype, and 90% vs. 10% using the broader phenotype which included social or cognitive deficits (Bailey et al., 1995). Across twin studies of autism, the difference between MZ and DZ concordance rates is sizable, averaging roughly 10:1 (Pennington, 2009). This rate is greater than that for other psychiatric disorders such as depression, bipolar disorder, and schizophrenia (between 2:1 and 4:1), indicating a high heritability for autism (Pennington, 2009). On the other hand, there was great variability in IQ and clinical behaviors in the 16 MZ pairs concordant for autism in the (Bailey et al., 1995) study. In other words, there was no more similarity for these traits within MZ pairs than that between individuals picked at random from different MZ pairs who also had autism. As interpreted by Pennington (2009), this finding suggests that although autism is heritable, the genes may not dictate the exact phenotype. Nonadditive interaction among genes (epistasis) and nonshared environmental influences likely contribute to these differences in phenotypes. Further, the large disparity between the MZ and the DZ concordance rates has been attributed to epistasis (Pennington, 2009). Alternatively, rare gene variants causing a common disorder (autism), as described below, could contribute to this large MZ/DZ discordance.

Familiarity

Individuals with autism rarely marry and have children so that vertical transmission of the diagnosis from parent to child is rarely observed (Pennington, 2009). However, genetic transmission is still possible as parents can transmit genetic risk factors without having the diagnosis themselves. Family studies (cited in Geschwind & Konopka, 2009; Pennington, 2009) suggest that the risk of autism is 20–60% higher in siblings compared to the incidence of autism in the general population, and to that of other psychiatric disorders. Several studies have shown that a broad autism phenotype is transmitted in families of individuals with autism (Piven, 1999; Rutter, 2000). For example, first-degree relatives of individuals with autism were shown to be shy, aloof, and have problematic pragmatic language (Rutter, 2000). This pattern is consistent with the segregation of sub-threshold traits within these families (Abrahams & Geschwind, 2008).

Gender Differences

Finally, autism affects more boys than girls (4:1), a finding which has remained constant since Kanner’s first description of autism in 1945, and despite the increasing incidence of this diagnosis. The predominantly male ratio has been attributed to an abnormality on the X chromosome (discussed below), or to sex linkage or genomic imprinting (Lintas & Persico, 2009; Marco & Skuse, 2006). Sex linkage involves a gene on the X chromosome transmitted from the mother to the son. As the son has only one X chromosome, this gene would be expressed. Since the mother’s daughter has two copies of the X chromosome – one from her father and one from her mother – the daughter likely would not express the abnormal phenotype. The most well-known example of a sex-linked disorder is hemophilia, which is on the X chromosome. Genomic imprinting, on the other hand, is an epigenetic phenomenon wherein chemical modification of DNA that does not alter the basic DNA sequence or modification of the DNA-associated histone proteins determines whether the maternal or the paternal copy of a specific gene is expressed. Genomic imprinting has been determined for Prader Willi and Angelman’s syndrome.

Finally, increased risk for autism has been identified in the offspring of older fathers (Reichenberg et al., 2006). Therefore, the gender and age of the parent may confer risk for offspring with autism.

Genetic Models of Autism

The conclusion drawn from indirect evidence is that autism is the most heritable and familial neurodevelopmental disorder (Pennington, 2009). With rare exceptions, however, autism does not appear to be the action of a single gene inherited in a strictly Mendelian pattern (autosomal dominant, recessive, or X-linked; Gupta & State, 2007; O’Roak & State, 2008). Rather, there are reports of multiple, distinctly rare changes in the genetic code in small subsets of individuals that cause or contribute to autism. There may be multiple gene variants – “a conspiracy of multiple genes” (Gupta & State, 2007) – that converge leading to a given phenotype.

Despite the indirect evidence for heritability and recent genetic technological advances, a genetic cause can be attributed to only 10–20% of all cases (reviewed in Abrahams & Geschwind, 2008), with a recent report suggesting a genetic cause can be uncovered in up to 40% of cases (Schaefer & Lutz, 2006). Further, Abraham and Geschwind (2008) state that no single genetic cause accounts for more than 1–2% of cases – similar to what is seen in mental retardation, another condition without a single genetic cause. While numerous studies identifying candidate genes or makers have been reported, very few studies have been replicated (Losh et al., 2008). Reasons to explain why candidates have not been agreed upon include the initial lack of uniform diagnostic criteria (strict vs. broad definition), limited power, varying methodology (Losh et al., 2008), and neglect of epigenetic factors modeling the disorder (Lahiri, Maloney, & Zawia, 2009). Further, as with other conditions with a strong heritable component, it appears that different genes may contribute to distinct components of the condition which gives rise to the full disorder through concerted actions (Losh et al., 2008; Pickles et al., 1995). Consequently, it has been said that linkage technology has not “found the autism gene,” but rather it demonstrated that more powerful technology is necessary to explain the multiple genes associated with autism (Abrahams & Geschwind, 2008). No one knows just how heterogeneous the syndrome is likely to be, that is, how many genes or regions of DNA (loci) may contribute either within a single individual or among the entire group of affected individuals. Some of the chromosomal regions and genes that have been associated with autism are summarized in Table 6.1 and will be addressed herein.

Table 6.1 The polygenetics of autism, selected chromosomal regions, and genes
Full size table

In addition, it is important to distinguish between locus heterogeneity, which refers to a variation at many different genes or loci resulting in a similar phenotype, and allelic heterogeneity, which refers to different variations or mutations at the same locus leading to an identical or overlapping clinical picture. Accumulating evidence suggest that both play a role in autism (O’Roak & State, 2008).

Recently, a rare-variant common disease model has been introduced (O’Roak & State, 2008). In this model, rare genes explain the common disease of autism. This makes sense in the Darwinian tradition that a deleterious change in the human genome leads to reduced fitness and that this would not be likely to propagate within a population. Autism fits the rare gene-common disease model as it begins early in life, impairs social interactions, and is associated with mental retardation which impacts reproductive fitness (O’Roak & State, 2008). The large difference seen between monozygotic and dizygotic concordance rates is consistent with de novo (rare gene) mutation. It appears that simplex families demonstrate de novo mutations more frequently – with one family member with autism (Sebat et al., 2007) – whereas multiplex families demonstrate transmitted variants (Bakkaloglu et al., 2008). The search for rare variants has led to the study of families that are genetically isolated, with shared ancestry and prone to consanguinity (Morrow et al., 2008; Strauss et al., 2006) to locate recessive alleles. Finally, rare sequence variations have been detected, thanks to improved technology such as CNV analysis. Indeed within the last 5–10 years, 20 bona fide risk genes have been identified due to the improved technology developed after the linkage method (Abrahams & Geschwind, 2008). Several of these risk genes will be discussed in the following sections.

Genetic Research

Abrahams and Geschwind (2008) write that two views underlie genetic autism research. The first is that rare variants involve an abnormality in a single molecule causing the clinical pathology in autism. Mendelian inheritance (autosomal dominant vs. recessive) takes this form. Technology compatible with this approach includes cytogenetics (including karyotyping and FISH), gene association studies (analysis of genes and protein system from less complex genetic syndromes similar to autism such as Rett and fragile X syndromes), linkage studies (including genome screens in affected sibling pairs), microarray technology, and CNV analysis. The second view involves the study of independent hereditary endophenotypes representing the core features of autism. An endophenotype is an operationally defined behavioral feature such as age of first spoken word. Although there is a general consensus regarding endophenotypes commonly seen in autism, the behaviors may not be specific to autism, as delay in language acquisition accompanies many developmental disorders. Ideally, endophenotypes represent discreet behaviors which do not overlap with other behaviors from the autism core domains (i.e., language, socialization, and rigid behavior). For example, poor eye contact, representing a deficit with socialization, is construed as being independent of repetitive behaviors, thought to represent the rigid behavior domain. Genetic study of these discreet endophenotypes, largely undertaken via linkage studies, narrows the scope of analysis (Losh et al., 2008; O’Roak & State, 2008).

Cytogenetics: Rare Mutations

Reports of chromosomal abnormalities detected by karyotyping first demonstrated that rare variants may contribute to autism (Vorstman et al., 2006). Estimates of chromosomal abnormalities in autism range from 6 to 40% (Abrahams & Geschwind, 2008; Marshall et al., 2008; Pennington, 2009; Schaefer & Lutz, 2006) so that genetic workup is now being recommended for all children diagnosed with autism (Pennington, 2009; Schaefer & Mendelsohn, 2008).

Chromosome 15q11-13

Chromosomal studies have produced a great interest in chromosome 15q11-13, the site of the most frequent chromosomal anomaly seen in autism (reviewed in Veenstra-VanderWeele & Cook, 2004). The maternally derived duplication of this region involves an imprinting mechanism. Maternal interstitial duplication or supernumerary inverted duplication of 15q11-13 is seen in 1–3% of patients (Cook et al., 1990; Schroer et al., 1998). Clinical features of chromosomal derangement in this region include mental retardation, motor impairment, seizure disorder, and impairment in communication, in some but not all with autism or attention-deficit hyperactivity disorder (ADHD; Cook et al., 1997; Schroer et al., 1998).

The duplication of 15q11-13 seen in some cases of autism is the opposite of deletions from the same region seen in Angelman’s syndrome (if inherited from the mother) or Prader Willi syndrome (if inherited from the father). Angelman’s syndrome, known as the “happy puppet” phenotype, involves mental retardation, epilepsy, ataxia, lack of speech, predominant laughing and smiling, and a high rate of autism. Prader Willi syndrome associated with mental retardation and hyperphagia is only occasionally associated with autism. Presently, there is ongoing research regarding how duplications and deletions in this gene region lead to an increased risk of autism (Veenstra-VanderWeele & Cook, 2004). This region, however, has not been identified as one of interest in whole genome searches, perhaps due to its rarity (Pennington, 2009). Still, FISH analysis of chromosome 15q11-13 is often performed in evaluation of children with autism. Recently, FISH is being replaced by CGH due to improved detection of abnormality within the 15q11-13 region.

Fragile X Syndrome

Cytogenetic approaches provided the first evidence for an autism gene 40 years ago when Lubs (1969) identified an abnormal or “fragile” site on the long arm of chromosome X in four males with mental retardation, leading to the recognition of fragile X syndrome (FXS). This syndrome, associated with mental retardation and autistic features, is more severely expressed in males. FXS is caused by a deficiency of the fragile X mental retardation protein (FMRP), resulting from little or none of the disease gene fragile X mental retardation 1 (FMR1) mRNA. The FMR1 gene mutation consisting of expanded CGG repeats of > 200 at chromosome site Xq27.3 is considered the origin of FXS. Autism, using the broad definition, has been reported in up to 30% of males with FXS, and FXS can be found in as many as 7–8% of individuals with autism (Muhle, Trentacoste, & Rapin, 2004). Later studies, using DNA measures of the fragile X mutation rather than cytogenetics and strict autism criteria, have found a smaller association between FXS and autism (Pennington, 2009). Two investigations, however, which studied carefully controlled groups of FXS-negative and FXS-positive males matched on intelligence, found higher rates of the following autistic symptoms in FXS males: gaze avoidance and hand flapping (Einfeld, Molony, & Hall, 1989) and stereotypic movements (including hand flapping, rocking, and hitting, scratching, or rubbing their own bodies), echolalia, gaze avoidance, and ritualistic behaviors (Maes, Fryns, Van Walleghem, & Van den Berghe, 1993). Studies of endophenotype behaviors rather than the strict autism criteria are more likely to uncover robust similarities between FXS and autism.

Exploring similarities between rare sub-groups of patients with a known disorder and those with a more common disorder (autism) provides a window into the shared biology between the disorders. FMRP has been shown to interact with multiple transcripts in repressing metabotropic glutamate receptor-5 (mGluR5) signaling activity which regulates long-term depression (LTD) associated with synaptic elimination. Without FMRP acting as a “brake,” mGluR–LTD is enhanced (Bear, Huber, & Warren, 2004). This favors an anabolic state which could contribute to the key features of FXS such as epilepsy, cognitive impairment, loss of motor coordination, and increased density of thin, long, dendritic spines in neurons (Bear, 2005). These observations have invited drug trials of glutamate receptor antagonists to reduce the expression of mGluR5 (Dolen et al., 2007) in individuals with FXS and in individuals with autism.

Investigation of Candidate Genes Contributing to Biochemical Pathways

Other genes contributing to medical conditions such as Alzheimer’s disease are under investigation to determine if they, too, share biology with autism and FXS. The discovery of fragile X ataxia syndrome (FXTAS) provides a precedent to study the relationship of developmental conditions across the life span within and between families. FXTAS is an adult-onset ataxia/dementia syndrome found in older family members of individuals with FXS (Hagerman, 2002; Hagerman et al., 2001). In contrast to what is found in FXS, FXTAS occurs with a rise in FMR1 mRNA level (Hagerman, 2006), fewer than 200 repeats (Basehore & Friez, 2009), and hypomethylation (Berry-Kravis, Potanas, Weinberg, Zhou, & Goetz, 2005) in the FMR1 CGG section on the X chromosome. This association of two clinically divergent disorders, regulated by the same gene produced in different doses, sets the stage for the investigation of the shared biology between autism, FXS, and other genes related to Alzheimer’s disease such as amyloid precursor protein (APP).

Amyloid Precursor Protein

APP, which is encoded by a gene on chromosome 21, is a large (695–770 amino acid) glycoprotein produced in several central nervous system (CNS) cell types including microglia, astrocytes, oligodendrocytes, and neurons. After protein processing, mature APP is axonally transported and can be released from axon terminals in response to electrical activity. APP is believed to play an important role in neuronal maturation and in synaptogenesis as reviewed by Lahiri, Farlow, Greig, Giacobini, and Schneider (2003). APP is of great interest because processing products of APP can include the insoluble 40–42-amino acid amyloid β-peptide (Aβ-40 and Aβ-42, respectively), the principal component of the cerebral plaques associated with memory and cognitive decline in Alzheimer’s disease (Alley et al., 2010). Recent research linking APP to autism illustrates how association studies enable the generation of new hypotheses about the biology of autism and ultimately advance our understanding of this disorder. Recently, we found evidence to support an association between autism and one of the APP pathways established for Alzheimer’s disease. In contrast to the upregulation of the amyloidogenic pathway as seen in Alzheimer’s disease, we found evidence that there may be an upregulation of the nonamyloidogenic (α-secretase amyloid precursor protein, sAPPα) pathway in a small sample of children with severe autism associated with self-injurious and aggressive behavior. That is, children with severe autism expressed total sAPP (representing the combined amyloidogenic and nonamyloidogenic pathways) at two or more times the levels of children without autism and up to four times the levels of children with mild autism (Sokol et al., 2006). Overall, there was a trend toward higher sAPPα within the children with autism. One of the severely autistic children in this study had FXS. High levels of sAPPα have also been found in a sample of children with autism from an independent laboratory (Bailey et al., 2008). High levels of sAPPα imply an increased α-secretase pathway in autism (anabolic), opposite to what is seen in Alzheimer’s disease. This would be consistent with the brain overgrowth hypothesis attributed to autism, in contrast to the brain atrophy seen with the deposition of amyloid plaque in Alzheimer’s disease. Coincidentally, via animal studies, Westmark and Malter (2007) concluded that FMRP binds to and regulates translation of APP mRNA through mGLuR5, providing a potential link between neuronal proteins associated with AD and FXS. By way of mGluR5, FMRP provides the “brake,” which if unchecked in conditions such as FXS would favor high levels of APP. Further, high levels of APP have been found in another study of FMR1 knockout mice (D’Agata et al., 2002). These findings have led to enticing speculation that the APP gene may influence several neurodevelopmental disorders across the life span.

Chromosome 7q

Both chromosomal inversions and translocations have been reported near 7q31 in boys with autism. One of these translocation breakpoints identified the deranged gene as RAY1/ST7 (FAM4A1), a putative tumor suppressor gene (Vincent et al., 2002), and work in this region is ongoing. Mutations leading to amino acid changes have been found on the WNT2 gene on 7q31 in two families with autism, and one affected parent transmitting the mutation to two affected children (Li et al., 2004; Wassink, Brzustowicz, Bartlett, & Szatmari, 2004). The FOXP2 gene on 7q31 was found to be disrupted in one family with an autosomal dominant form of specific language impairment (SLI) (Lai, Fisher, Hurst, Vargha-Khadem, & Monaco, 2001), and this was replicated in a larger study of SLI (O’Brien, Zhang, Nishimura, Tomblin, & Murray, 2003), although initial studies failed to associate this with autism (Newbury et al., 2002; Wassink et al., 2001). Recently, FOXP2 differences in gene expression were found between chimp and human cell cultures (Konopka et al., 2009). Compared to chimps, in human culture, the FOXP2 gene affected transcription upregulation in 61 genes and downregulation in 55 genes. Further, the FOXP2 gene was shown to regulate downstream effects involving cerebellar motor function, craniofacial formation, and cartilage and connective tissue formation required for expressive language. As discussed in section “Endophenotype,” there is a suspected association between the chromosome 7q region and speech abnormality seen in autism.

The RELN gene maps to chromosome 7q22. This gene encodes a protein that controls intercellular interactions involved in neuronal migration and positioning in brain development. A large polymorphic trinucleotide repeat in the 5′-YTR of the RELN gene has been implicated in autism in several studies (Ashley-Koch et al., 2007; Persico et al., 2002; Zhang et al., 2002). Further support for this candidate gene comes from the RELN mouse model which carries a large deletion in RELN and shows atypical cortical organization similar to cytoarchitectural pathological anomalies reported in the brains of individuals with autism (Bailey et al., 2008).

Phosphatase and Tensin Homolog Gene (PTEN)

PTEN is a tumor suppressor gene located on chromosome 10q23. This gene influences the cell–cell arrest cycle and apoptosis or programmed cell death (Lintas & Persico, 2009). PTEN inactivation causes excessive growth of dendrites and axons, with an increased number of synapses (Kwon et al., 2006). Mutations in PTEN cause Cowden’s syndrome (macrocephaly, hamartomas, and autism). This gene has been of interest to autism researchers as macrocephaly has been considered to be one of the “most widely replicated biological findings in autism” affecting up to 20% of all children with the condition (McCaffery & Deutsch, 2005). Butler et al. (2005) examined the PTEN gene in 18 individuals with autism and macrocephaly. They discovered three with PTEN missense mutations. Others have found PTEN mutations in macrocephalic patients with autism (Herman et al., 2007a, 2007b).

Other Cytogenetic Regions

Cytogenic abnormalities have included deletions and duplications involving 2q37, 22 q13, 22q11, 17p11, 16p11.2,18q, Xp, and sex chromosome aneuploidies (47, XYY and 45, X/46, and XY mosaicism) (Costa e Silva, 2008; Mendelsohn & Schaefer, 2008; Schaefer & Mendelsohn, 2008; Sykes & Lamb, 2007). These genetic findings have generated interest in testing the association of a number of candidate genes in these regions via linkage and animal studies.

Autism and Synapse Formation (Synaptogenesis) Overview

Accumulating evidence points to the involvement of three genes (neuroligin, SHANK3, and neurexin) in the synapse formation of glutamate neurons. Glutamate is an excitatory neurotransmitter and aberrant glutamate function has long been suspected to contribute to autism. Neuroligins induce presynaptic differentiation in glutamate axons. SHANK3 encodes for a postsynaptic scaffolding protein which regulates the structural organization of dendritic spines in neurons. Consequently, SHANK3 is a binding partner of neuroligins. Neurexin induces glutamate postsynaptic differentiation in contacting dendrites. Altogether, these genes appear to contribute to glutamatergic synapse formation.

Neuroligin Genes (NLGN3 and NLGN4)

Neuronal cell adhesion is important in the development of the nervous system, contributing to axonal guidance, synaptic formation and plasticity, and neuronal–glial interactions (Glessner et al., 2009; Lien, Klezovitch, & Vasioukhin, 2006). Derangement in cell adhesion could contribute to migrational abnormalities including brain overgrowth. Neuroligins are cell-adhesion molecules localized postsynaptically at both glutamatergic (NLGN1, NLGN3, NLGN4, NLGN4Y) and γ-aminobutyric acidergic (NLGN2) synapses (Lintas & Persico, 2009). Neuroligins trigger the formation of functional presynaptic structures in contacting axons. As mentioned above, they interact with postsynaptic scaffolding proteins such as SHANK3 (see below). The NLGN3 and NLGN4 genes are located at chromosomes Xq13 and Xq22, 33, respectively, and mutations here have been associated with autism Jamain et al. (2003). Extensive genetic screens conducted by several research groups have confirmed the low frequency of neuroligin mutations among individuals with autism (Lintas & Persico, 2009). For example, Jamain et al. (2003) found a frameshift mutation in NLGN4 and a missense mutation in NLGN3 in two unrelated Swedish families, inherited from unaffected mothers. Laumonnier et al. (2004) reported a frameshift mutation in NLGN4 in 13 affected male members of a single pedigree. Lawson-Yuen, Saldivar, Sommer, and Picker (2008) found a deletion of exons 4–6 of NLGN4 in a boy with autism and in his brother with Tourette syndrome whose mother showed psychiatric problems and also carried the mutation. Neuroligin mutation carriers, however, display a variety of syndromes, such as X-linked mental retardation without autism (Laumonnier et al., 2004), Asperger’s syndrome, autistic disorder of variable severity, and PDD-NOS (Yan et al., 2005). The symptoms may be slow, or abrupt and associated with regression. Despite intensive investigation, the low frequency of neuroligin mutations and the lack of similar phenotypes have led to the recommendation that neuroligins should not be included in widespread screens for individuals with nonsyndromic autism (Lintas & Persico, 2009).

SH3 Multiple Ankyrin Repeat Domains 3 Gene (SHANK3)

The SHANK3 gene is located on chromosome 22q13.3 which encodes for a scaffolding protein found in the postsynaptic density complex of excitatory synapses binding to neuroligins. Recently, two studies have reported a correlation between mutations affecting SHANK3 and an autism phenotype characterized by severe verbal and social deficits (Durand et al., 2007; Moessner et al., 2007). Of the seven patients reported with the SHANK3 gene mutations, five were deletions, one was a missense, and another a frameshift mutation. In addition, rare SHANK3 variations were present in the autism group, but not in the control group. These variations were inherited from healthy parents and they were present in unaffected siblings, perhaps demonstrating incomplete penetrance. Another interesting observation is that in both studies (Durand et al., 2007; Moessner et al., 2007), the SHANK3 deletion was inherited via a paternal balanced translocation. Further, in both studies, siblings of the probands with SHANK3 abnormalities had partial 22q13 trisomy that resulted in attention-deficit hyperactivity disorder (Moessner et al., 2007) and Asperger’s syndrome with early language development (Durand et al., 2007). Like neuroligins, SHANK3 mutations are very rare. However, because of the robust genotype–phenotype correlation reported in two studies (Durand et al., 2007; Moessner et al., 2007), it has been recommended that children with severe language and social impairment obtain SHANK3 mutation screening (Lintas & Persico, 2009).

Neurexin Genes

Presynaptic neurexins influence postsynaptic differentiation in contacting dendrites by interactions with postsynaptic neuroligins. Three neurexin genes (NRXN1, NRXN2, and NRXN3) are located on chromosome loci 2q32, 11q13, and 14q24.3-q31.1, respectively. Again, neurexin mutations are very rare. For example, two missense mutations were present in 4 of 264 (1.5%) individuals with autism compared to none in 729 controls (Feng et al., 2006). However, in this study, missense mutations also occurred in first-degree relatives who displayed heterogeneous phenotypes such as hyperactivity, depression, and learning problems. Incomplete penetrance may explain these findings, or autism may be caused by neurexin acting synergistically with other susceptibility genes. Neurexin-1α is being intensely studied in mice as deletion in this molecule resulted in increased repetitive behavior, whereas social behavior was relatively intact (Etherton, Blaiss, Powell, & Sudhof, 2009). This implies an animal model for a discreet feature of autistic behavior (endotype), as discussed below.

Contactin-associated protein-like 2 (CNTNAP2) is a member of the neurexin superfamily which involves a recessive frameshift mutation on chromosome 7q35. It is one of the largest genes of the human genome and encodes CASPR2, a transmembrane scaffolding protein. CNTNAP2 has been associated with cortical dysplasia–focal epilepsy in an Old Order Amish community (Strauss et al., 2006). Autism was present in up to 67% of these individuals. We recently reported CNTNAP2 in an Amish girl with epilepsy and autism who also showed hepatosplenomegaly (Jackman, Horn, Molleston, & Sokol, 2009). CNTNAP2 has been associated with an autism language phenotype in large studies of non-Amish individuals (Alarcon et al., 2008). Further, stage two of this investigation showed that CNTNAP2 was expressed in the language centers (frontal and anterior temporal lobes) of fetal brain (Alarcon et al., 2008). It has been recommended that Amish children presenting with autism should be tested for CNTNAP2 gene mutation (Strauss, personal communication). This work again supports a language function associated with chromosome 7.

The Methyl-CpG-Binding Protein 2 Gene

Methyl-CpG-binding protein 2 (MeCP2) works as a transcriptional repressor within gene-promoting regions involved in chromatin repression. This gene is located on chromosome Xq28 and shows mutation in 80% of females with Rett syndrome (pervasive developmental disorder, acquired microcephaly, epilepsy, and loss of hand function). This gene has been studied in children with autism, and MeCP2 mutations are considered to be rare (0.8–1.3%) in females with autism. Interestingly, autism and Rett syndrome share some similarities at the phenotypic and pathogenic levels, and both disorders were proposed to result from disruption of postnatal or experience-dependent synaptic plasticity (Zoghbi, 2003). Among the Rett mutations reported are a frameshift mutation, a nonsense mutation, and additional introns (Lintas & Persico, 2009). Schaefer and Mendelsohn (2008) observe that MeCP2 has not been associated with idiopathic autism in males so that this gene test is recommended only for females with autism.

Linkage Studies

Linkage studies involve determining whether the transmission of a chromosomal segment from one generation to another coincides with the presence of the phenotype of interest (Gupta & State, 2007). Linkage can be utilized in Mendelian inherited conditions and can also be used to study complex conditions such as in autism when Mendelian inheritance is unlikely, and there is no hypotheses regarding the specific nature of transmission.

Linkage studies can be grouped into two types – the conventional study of a chromosomal region of interest in affected sibling–pairs from multiplex families and the more recent genome-wide linkage analysis in which every chromosome is evaluated simultaneously. In the sib–pair study method, siblings with autism are evaluated to determine whether they share any regions of the genome more frequently than would be expected by chance.

Genomic wide association studies compare genetic risk factors in the form of specific genetic markers in cases and controls. These markers are distributed within the entire genome rather than limited to specific gene regions such as in the sib–pair method. This enables a more unbiased ascertainment of regions of interest (Losh et al., 2008). Several genome-wide scans of individuals with autism have been reported and evidence in favor of linkage has been determined for the majority of chromosomes (Gupta & State, 2007). The trouble with these studies is that the evidence has not reached statistical significance and there is lack of replication for many of these findings. Statistical significance is calculated by a logarithm of the odds (LOD) score. This score represents the logarithm of the likelihood ratio of observing the data under a model of linkage to observing the data under a model of free recombination (no linkage). An LOD score of 3.5 in a sib–pair analysis is considered to be a significant linkage (Lander & Kruglyak, 1995). Suggestive linkage is an LOD score of 2.2, and a highly significant LOD score is 5.4. For replication studies, the replication threshold is considered to be 1.4 (Lander & Kruglyak, 1995). By chance, one would expect to see a suggestive linkage peak once every genome scan or a significant peak once every 20 scans (Gupta & State, 2007).

Most linkage studies have identified linkage regions reaching the level of suggestive linkage at best (Freitag, 2007). Despite large increases in the size of patient cohorts, linkage signals have not increased significantly with sample size (Abrahams & Geschwind, 2008). Only three loci (2q, 7q, and 17q11-17) have been replicated for nonsyndromic autism and are considered to have genome-wide significance. Genomic-wide screens have engendered great interest in chromosome 7q with distinctive peaks involving two distinct regions: 7q21-22 and 7q32-36 (International Molecular Genetic Study of Autism Consortium-IMGSAC, 1998; Collaborative Linkage Study of Autism, 2001). As chromosome 7q has been discussed in section “Cytogenetics: Rare Mutations,” chromosome 2 and then 17q11 will follow the description of how linkage studies led to the discovery of the gene loci for a syndromic form of autism: tuberous sclerosis complex (TSC). The sibling–pair from multiplex family design has been used to study autism within many genetic loci, including those involved in TSC; the genome-wide linkage analysis has detected linkage on chromosomes 7q, 2q, 17q11, and novel loci.

Tuberous Sclerosis Complex

Tuberous sclerosis complex (TSC) is a neurodevelopmental disorder characterized by cognitive delay, epilepsy, and neurocutaneous growths including hamartomas (i.e., tubers) within the central nervous system and other organs. Up to 60% of individuals with TSC have autism. The first suspicion of a chromosomal abnormality associated with TSC originated from a linkage study of 26 protein markers within 19 multigenerational families affected with TSC (Fryer et al., 1987). In eight of the families, ABO blood group gene mapped to chromosome 9q34.3. Many groups corroborated these results using larger numbers of families (Au, Williams, Gambello, & Northrup, 2004). These linkage studies established the TSC1 gene, later discovered to encode hamartin, a growth suppressor protein.

Another TSC gene site was discovered via linkage studies on chromosome 16p13.3, known as TSC2. This gene produces the tumor suppressor protein tuberin. Further evidence showed that hamartin works together with tuberin in several cell-signaling pathways including a growth and translation regulatory pathway (P13K/PKB), a cell adhesion/migration/protein transportation pathway [glycogen synthase kinase 3 (GSK3)/β-catenin/focal adhesion kinase/Ras-related homolog (Rho) pathway], and a cell growth and proliferation pathway [mitogen-activated protein kinase (MAPK)]. The tuberin–hamartin complex affects mTOR kinase activity (Au et al., 2004), promoting tumor growth. This discovery has led to the study of rapamycin, a drug used in organ transplant and cancer treatment, as a therapy for suppressing growth of tumors in TSC (Kenerson, Aicher, True, & Yeung, 2002).

Other growth-promoting genes such as PTEN have been linked to the tuberin–hamartin gene complex as contributors to the general overgrowth in TSC. As previously discussed, PTEN is a gene of interest in autism.

Chromosome 2q

Buxbaum et al. (2007), together with the Seaver Autism Research Center (SARS), reported a two-point dominant LOD score of 2.25 on chromosome 2q in 35 affected sibling pairs. In a second-stage screening which employed 60 families with autism probands, the strongest linkage was at this same location. The IMGSAC study showed a strong linkage to 7q as described above, but also found linkage to 2q (IMGSAC, 1998), with a more recent study also providing evidence for linkage to 2q (IMGSAC, 2001b).

Chromosome 17q

Early recognition of the need for large patient cohorts and substantial genetic heterogeneity led to the establishment of the autism genetic resource exchange (AGRE) composed of over 500 families. AGRE is a publically available resource of phenotypic data and biomaterials. Genomic scan linkage analysis was performed on 109 autistic sibling pairs from 91 AGRE families together with analysis of those pairs from an independent sample of 345 families from the same AGRE cohort (Cantor et al., 2005). When families with autism were stratified into only those with affected males, there was significant linkage at 17q11-21 in both samples. The LOD score for this replicated work was at genome-wide level of significance (LOD score > 1.4). One positional candidate gene close to this region is the serotonin transporter gene (SLC6A4) which codes for a protein controlling the reuptake of serotonin in the synapse. This was exciting as elevated levels of serotonin have been determined in 25–30% of cases of autism (Cook et al., 1990). Indeed, relationship between the SLC6A4 site and a repeat polymorphic region in its promoter region (5HTTLPR) recently has been independently investigated (Losh et al., 2008). The chromosome 17q locus, however, has not been uniformly observed in subsequent, large-scale studies (Schellenberg et al., 2006) which may be due to phenotypic heterogeneity.

Novel Loci for Autism

A very recent genome-wide linkage mapping study in a sample of 1,031 multiplex autism families (Weiss, Arking, Daly, & Chakravarti, 2009) identified significant linkage on chromosome 20p13 and suggestive linkage on chromosome 6q27. In this study, no associations meeting criteria for genome-wide significance were found, suggesting there are not many common loci of moderate to large effect size. However, replication data revealed a novel SNP locus on chromosome 5p15. This region was adjacent to SEMA5A, a member of the semaphoring axonal guidance protein family which has been shown to be downregulated in transformed B lymphocytes from autism samples (Weiss et al., 2009). The authors further demonstrated lower SEMA5A gene expression in autism brain tissue. This finding is in keeping with the suspected derangement in the migration of cortical neurons during embryogenesis in autism.

Microarray Technology

Microarray technology is transforming the identification of chromosome duplications and deletions (Gupta & State, 2007). This new technology, known as high-density, oligonucleotide-based array comparative genomic hybridization (aCGH), is now available for widespread use. This technique uses patient DNA and control DNA, each labeled with a fluorescent tag (red or green). Equal amounts of DNA from patient and control are hybridized to known regions of the human genome pre-arrayed on a slide. If patient and control have equal copy numbers at a given locus, the color turns yellow, representing an equal measure of DNA. If the patient has lost (deleted) a locus, only the control color is visualized. If the patient has an extra copy at a locus (duplication), the patient color predominates. The aCGH probe may be enriched for known genes or specific chromosomal regions for known syndromes, or distributed evenly across the whole genome. This new technique is now available at all major medical centers and through Signature Genomic Laboratories (http://www.signaturegenomics.com).

Single-nucleotide polymorphism (SNP) analysis probes thousands of SNPs and provides data about copy number and genotype (Li & Andersson, 2009). The genotype can be used to study uniparental disomy (UPI) seen in imprinting disorders and consanguinity. This technique is better at focused study of specific gene regions instead of the whole genome. Additionally, SNP analysis uses pre-established laboratory standards rather than intraexperimental controls. Both SNP analysis and aCGH map produce copy number variation, and it is common for labs to first perform aCGH and follow up with SNP in specific regions of interest.

Today, CNV from several thousand nucleotides can be identified, greatly improving upon the sensitivity of conventional cytogenetics. Sebat et al. (2007) showed that CNVs were present in 10% of affected individuals from single-incidence autism families (i.e., sporadic cases), contrasting with substantially lower rates observed in controls (1%) and autism cases from multiplex families (3%). This has led to the general expectation that de novo CNVs are more likely to be found in sporadic (and as it turns out, dysmorphic) cases. Jacquemont et al., 2006 reported CNV rearrangements in 8 (27.5%) of 29 patients with syndromic autism (including facial dysmorphism, limb or visceral malformations, and growth abnormalities). There were no reoccurrence or overlap in these variants for the eight children. Chromosome 11q12-p13 and neurexins were implicated in a linkage study using SNP–CNV in 1,181 families with at least two affected individuals through the Autism Genome Project (Szatmari et al., 2007).

Microarray technology, however, comes with the realization that typically developing individuals have more structural variants than previously imagined (Sebat et al., 2007). There is growing concern that CNV irregularities may not be pathological and therefore not be a cause of autism (Tabor & Cho, 2007). These authors note that using diagnostic tools prematurely in a clinical context may be “unethical, either because of over-treatment, under-treatment, unwarranted labeling and stigmatization, or a false sense of security.” Certainly, further studies on large cohorts of children with the same deletion/duplication are necessary to enable clinical application of this technology.

Endophenotypes

Due to the phenotypic heterogeneity of autism and the lack of finding a specific gene, researchers have narrowed the scope of analysis to more pure, operationally defined behaviors/traits. Endophenotypes are behavioral, physiological, and/or neuropsychological markers that are present in both affected and unaffected individuals. Rather than searching for “autism genes,” endophenotype investigations search for smaller grouping of genes that contribute to discreet phenotypes (Losh et al., 2008). This approach allows for measurement of “dosage” of a trait and can be applied to affected as well as unaffected individuals.

Specific Language Impairment

Specific language impairment (SLI) is defined as the failure of normal development of language without hearing loss, mental retardation, or oral motor, neurological, or psychiatric impairment. This affects approximately 7% of children entering school (Tomblin et al., 1997). Individuals with SLI perform poorly on phonologically based tasks and many go on to develop dyslexia (Stothard, Snowling, Bishop, Chipchase, & Kaplan, 1998). There appears to be neurostructural association between SLI and autism (Herbert et al., 2002; 2004), and there is an associated finding of language delay in relatives of probands with autism (Wassink et al., 2004). Several investigations have found strong linkage of SLI to chromosome 13q21 (Bartlett et al., 2002; 2004) and to chromosomes 16q and 19q (SLI Consortium, 2002, 2004). This finding is noteworthy as it directly overlies the 16q locus linked to autism (Wassink et al., 2004). A specific language marker, age of first word, has shown significant linkage to chromosome 7 observed by five separate investigators (Losh et al., 2008). The 7q region has been the subject of intense investigation, as described above, and may be the loci associated with autism language.

Restrictive and Repetitive Behavior

The restricted and repetitive behavior (RRB) endophenotype in children with autism is receiving attention (Morgan, Wetherby, & Barber, 2008). RRB comprises one prong of the autism clinical triad (language deficit, social deficit, and RRB). Cuccaro et al. (2003) identified two factors underlying RRB as measured by the Autism Diagnostic Interview-Revised (ADI-R): repetitive sensory motor actions and resistance to change. This is of interest as a genetic linkage signal has been reported for resistance to change on chromosome 15 (Shao et al., 2003). It appears that the analysis of phenotypic homogeneous subtypes may be a powerful tool for mapping of candidate genes in complex traits such as autism.

Clinical Genetics Evaluation

The recent marked increase in the incidence and awareness of autism has resulted in an increase in the number of children sent for diagnostic evaluation. The general consensus within the literature is that genetic consultation should be conducted on all persons with the confirmed diagnosis of autism. While referral for genetic consult is often preferred, often the primary practitioner, pediatrician, and/or pediatric neurologist are in a position to conduct the initial evaluation. Further, clinical geneticists may not be available, particularly in underserved areas of the country, or a timely genetic evaluation may not be possible due to lengthy waiting lists. Therefore, the clinician taking care of a child with autism may wish to initiate the genetic evaluation.

Recently, sequential guidelines for clinical genetics evaluation in autism have been published by Schaefer, Mendelsohn, and the Professional Practice and Guidelines Committee (2008) (see Box 6.5). These evidence-based guidelines have evolved from an original retrospective study of patients referred for clinical genetics evaluation between the years 2002 and 2004 at the University of Nebraska Medical Center (Schaefer & Lutz, 2006). The guidelines are dynamic rather than static and have been updated, for example, as aCGH has become more available. The stepwise approach was designed to balance cost with the expected yield of the tests. Further, there is a pyramidal effect so that “earlier tiers have a greater expected diagnostic yield, lower invasiveness of testing, better potential of intervention, and easier overall practicality” (Schaefer & Mendelsohn, 2008). Summarizing the diagnostic yields expected for the following investigations (high-resolution chromosome studies, 5%; aCGH – beyond that detected by chromosome analysis – 10%; fragile X, 5%; MECP2, 5% women only; PTEN, 3% if head circumference >2.5 SD; other, 10%), it was predicted that utilization of these guidelines would lead to diagnosis in 40% of cases (Schaefer, Mendelsohn, and the Professional Practice and Guidelines Committee (2008). Finally, this stepwise approach has met the approval of third-party payers and families (Schaefer & Lutz, 2006).

 

Pre-evaluation

This initial step includes confirming the diagnosis of autism using objective (DSM-IV) criteria and/or standardized objective measures such as those discussed in Chapter 15. An audiogram is obtained to rule out hearing loss and an electroencephalogram (EEG) is obtained if there is a clinical suspicion of seizures. Cognitive testing, when appropriate, can determine mental retardation. Finally, verifying the results of the newborn screen can help rule out rubella and PKU.

Tier 1

Initial assessment involves a standard clinical genetic history and physical exam to identify known syndromes or associated conditions. Included would be a Wood’s lamp examination of the skin to help rule out neurocutaneous conditions such as TSC. Also, for suspected diagnoses, standard metabolic screening is performed to check for urine mucopolysaccharides and organic acids as well as serum lactate, amino acid, ammonia, and acylcarnitine profile. If not already performed, high-resolution chromosomal analysis and DNA for fragile X is sent.

Tier 2

If the studies in the first tier are unrevealing, the second tier checks for aCGH duplications and deletions. For patients with pigmentary abnormalities on exam but with a normal leukocyte karyotype, skin biopsy can be obtained to obtain a fibroblast karyotype. For females, MECP2 gene testing is obtained; for children with head circumference greater than 2.5 SD from the mean, PTEN gene testing is recommended.

Tier 3

Lower yield tests have been reserved for this level: brain magnetic resonance imaging and serum and urine uric acid. Further tests are outlined depending on the results (high or low) of the uric acid tests. We would add here that if a child has significant language impairment, a Shank3 mutation should be ruled out.

Finally, new susceptibility loci that can contribute to the autism phenotype are continually identified and catalogued in the Online Mendelian Inheritance in Man (OMIM) database (http://http//www.ncbi.nlm.nih.gov/sites/entrez). Using this search engine, a patient’s phenotype including dysmorphic features can be entered into the program to generate a list of possible genetic diagnoses.

Counseling

Most clinical geneticists work with genetic counselors who interpret the findings into recurrent risks for full siblings. The tiered clinical genetics evaluation should identify two groups of individuals: those with and those without an identifiable etiology for autism. For those without an identifiable etiology, counseling should be provided according to multifactorial inheritance (Schaefer & Mendelsohn, 2008). That is, 4% recurrence rate if the proband is a girl and 7% if the proband is a boy. If a second child has autism, a reasonable recurrence rate, based upon published reports, is 30%.