Have you ever been driving when suddenly your car’s “check engine” warning light comes on? You take the car to the mechanic, who tells you “more than a hundred different malfunctions can trigger that light.” Autism, we now know, is a bit like that, with potentially hundreds of genetic variations leading to the warning signs of the disorder.1 Unfortunately, determining exactly which of those many possibilities is the real culprit is not as simple as hooking your car’s central computer to a diagnostic tester. Rather, in autism many culprits may be acting at once, each contributing something different to the problems that characterize the disorder.
Autism: Not one disease
Autism spectrum disorders (ASDs) are a class of varied disorders typified by impairments in language development, social development, and narrow or restricted interests and/or repetitive behaviors. Most children with an ASD have normal intelligence.
A minority (about 35 percent) have intellectual disability, which is defined as an IQ below 70.2
A small minority of so-called autistic savants may be capable of remarkable memory feats or other capabilities.3
The prevalence of ASDs has increased 78 percent since 2002 according to the most recent estimates by the Centers for Disease Control and now stands at 1 in 88 children.4
ASDs are nearly five times more common in boys (1 in 54) than girls, and although all racial and ethnic groups are affected, ASDs are most often diagnosed in Caucasians.5
It is widely believed that at least some of the increase in ASD prevalence is due to improved, expanded, and earlier diagnosis.6,7 Cold or indifferent parenting has been discredited as the cause of autism, and current research focuses on more plausible biological mechanisms. Chief among these are explanations that seek to understand how brain development may be affected by unique differences in the genes carried by people with ASDs. Indeed, the genomics revolution has profoundly changed our understanding of autism and ASDs over the last decade, which has led not only to better knowledge of the disorder’s complexity but to an expanded list of possible treatment targets. ASD has, in effect, become a model for both the benefits and the challenges of complex traits—that is, traits not caused by the inheritance of single genes.
Figure 1: ASD Classification. Image source: Autism phenotypic classification. Devlin, B., & Scherer, S. W. (2012). Genetic architecture in autism spectrum disorder. Current opinion in genetics & development.
The challenges of diagnosis
As with psychiatric disorders, immune system disorders, and other diseases with varied and overlapping symptoms, ASDs can be difficult to diagnose. Typically, some sort of developmental delay will lead parents to consult with physicians or psychologists, who will base their diagnoses on observable behaviors as compared with “typical” milestones, such as age at first speech, ability to maintain eye contact, repetitive behaviors, etc.
Specific clinical criteria are defined in the Diagnostic and Statistical Manual of Mental Disorders, which is in its fifth edition (the so-called DSM-IV).8
The assessment of behaviors and characteristics are, to a large extent, subjective and depend on the skill and experience of the diagnosing clinician.
The criteria for Asperger’s syndrome overlap to a substantial degree with autistic disorder, and to a lesser degree with certain other developmental problems of childhood, such as Rett’s syndrome. In fact, the new edition of the Diagnostic and Statistical Manual of Mental Disorders (5th edition) will eliminate Asperger’s as a separate diagnosis and simply merge it with ASDs.
There are no definitive “biomarkers” for autism or ASD, such as molecules in the blood that can be detected by a simple medical test. However, laboratory tests and other diagnostics may be used to exclude alternative diagnoses, such as hearing loss, poisoning, or chromosomal abnormalities. Also, a number of known genetic disorders can produce symptoms that overlap with ASD, or ASD may occur in conjunction with them. For example, the vast majority (up to 90%) of patients with Potocki-Lupski syndrome, which is caused by a known mutation and includes multiple behavioral and other effects, also have an ASD.9
Advancing understanding through genomics
Genomics—the study of genetic variation across all genes, rather than in single genes, or, even more broadly, how genes operate at the level of whole organisms—has been made possible by new developments in technology. Chip technology adapted from the computer industry permits the study of thousands of genes at once. However, the biological principles underlying genomics are fundamentally genetic:
Most genes come in pairs that can be the same or different, called “alleles” (one copy from each parent), and the two copies “segregate” during the processes of sperm and egg production.
Different genes generally “assort” independently. This refers to the randomness by which different chromosomes separate, also during the production of gametes.
Mendelian disorders, such as cystic fibrosis or sickle cell, can be investigated using techniques that reveal mutations in one or both copies of a single gene. Genomics, in contrast, allows the study of complex traits and how they are affected by mutations in many genes.
Autism is known from twin studies to have a heritability as high as 90% or as little as 40%, meaning that 40 to 90% of the variation in autistic traits in a studied population is due to genetics.10,11 Part of this large uncertainty is due to differences in the diagnostic criteria used in the research studies (e.g., strict autism versus ASD). However, an important point is that not all of the variation is due to genetics. A sizable amount is attributable to the environment, and this influence is poorly understood. A new area of genomic research called epigenetics is starting to demonstrate how the environment can alter the activity of genes. Whereas the structure or sequence of a gene is fairly stable, the function of the gene varies over time and the course of development. It is by altering gene function through chemical modifications, known as epigenetic marking, that environment exerts its influence on genes while leaving the genes’ essential structure intact.
- Studies conducted on the brains of male suicide victims showed that those who had been abused as children had substantially lower activity in a gene involved in stress response than victims who had never been abused or people who had died of accidents, although the gene’s essential structure was the same.12
Complicating things further, it is likely that the manner in which an individual responds to the environment is itself directly or indirectly influenced by his or her genes, a phenomenon called gene-by-environment interaction. For example, pharmacogenomics is the study of how genetic differences affect both drug efficacy and drug side effects.
Knowing that genes are involved in autism and ASD is not the same as knowing which genes are involved or how. A major challenge in pinning down the precise genetic underpinnings of ASDs is related to the tremendous variation that characterizes the “phenotype” or specific traits of the disorder. For example, microdeletions (small losses of DNA) in the gene contactin-associated protein-like 2 (CNTNAP2) have been linked to language problems that are highly variable, with some cases showing relatively mild symptoms, such as delayed speech, while other cases with a similar deletion are much more severe (e.g., absent speech).13,14 In fact, a promising approach to finding more ASD genes involves studies structured to find genes associated with the specific symptoms of ASD or autism rather than with an overall diagnosis of “ASD” or “autism.”15 However, even here there will be complications that foil attempts to establish clear causation between newly found genes and ASD. Research suggests that genes involved in ASD also will be involved in other psychiatric conditions, for example, schizophrenia, bipolar disorder, and major depression.16,17
Thus far, the central message of genomics has been that variation is much more common than could have been imagined a decade ago. This has profound implications for the diagnosis of genetically influenced, complex diseases such as ASDs, and for ferreting out the underlying causes of those diseases. Variation (a broad term that includes the concept of ‘mutation’) comes in several varieties:
Small, single-nucleotide polymorphisms are single-base changes in the As, Ts, Cs, and Gs that make up the DNA in genes and are often abbreviated as SNPs.
Large or small insertions (extra chunks of DNA) or deletions (missing chunks of DNA) are often referred to as indels or structural variation.
Is genetic the same as inherited?
When most people think about a genetically influenced disorder, they assume this means inherited, but ASDs help illustrate an important distinction. A disease can be caused by a mutation in one or several genes yet be sporadic. In such cases, the mutations just happen—often as a result of mistakes during routine copying of DNA as sperm or eggs are produced. Such mistakes are often, but not always, repaired by the cell. If they escape repair, the resulting mutations are said to be de novo (or new) because they were not inherited from the parental generation, and they may cause disease in the offspring in ways that cannot easily be predicted. By contrast, mutations that lead to well-recognized Mendelian genetic disorders, such as sickle cell anemia, cystic fibrosis, or Huntington’s disease, are typically inherited from one or both parents and often lead to distinctive and predictable disease patterns in families.
- Roughly 15% or more of ASD cases are caused by de novo mutations.18
Advanced parental age appears to be a risk factor for ASD and may represent contributions of mutations that are both de novo and inherited (in the more traditional sense).
Fathers 40 years of age or older are roughly 40 percent more likely to have a child born with an ASD than fathers 25-29 years of age.19
There is also evidence that increased maternal age is a risk factor for ASD, with younger mothers conferring the least risk and older mothers conferring greater risks that are roughly proportional to age at conception.20
The accumulation of de novo mutations over time may help explain the age-associated risks for ASD as well as a number of other complex, genetic diseases such as schizophrenia. A study by Kong et al. (2012) looked at the entire genomes of 78 Icelandic trios (an affected child and both parents), estimated the rate at which SNP mutations occur, and concluded that increasing paternal age correlated with larger numbers of mutations.21
Rare versus common mutations
Interestingly, although de novo mutations are usually rare and affect seemingly random locations in the genome, recurrent de novo mutations that affect the same region in different individuals have been found as well, such as a deletion of a segment of DNA on chromosome 16 that involves about 30 genes. This mutation is sometimes present in persons without autism, but it is 100 times more common in those affected.22,23
O’Roak and colleagues (2011) not only found potentially causative de novo mutations in several genes, including FOXP1, which has been implicated in language development, they also found de novo mutations in combination with a rare inherited mutation.24 The latter finding supports the hypothesis that in many cases of ASD, perhaps most, more than one mutation may be required to explain the disorder.
Rare structural variation including duplications or deletions of DNA, also called copy number variation (or CNVs), may account for a significant amount of the unexplained genetic basis of ASDs and other neuropsychiatric diseases. CNVs larger than 50 kbp (or 50,000 base pairs) occur in roughly 10 percent of patients with ASD and are six times more common than in control patients without ASD.25
Types of genomic structural changes affecting segments of DNA. Image credit: Estivill X, Armengol L (2007) Copy Number Variants and Common Disorders: Filling the Gaps and Exploring Complexity in Genome-Wide Association Studies. PLoS Genet 3(10): e190.
In some studies, common, inherited mutations have been clearly implicated in ASDs, such as mutations in the CNTNAP2 gene mentioned above. In other studies, made possible only with modern genomic techniques, researchers have found a new way to look for recessive inheritance, one of the modes of inheritance identified by Gregor Mendel in the mid-19th century. Casey and colleagues (2012) looked for stretches of DNA on both chromosomes that were similar and more common in patients with ASD than in controls. In this way they identified a number of genes that may increase the risk for ASD only when both are mutated. This is a case of new genomic technologies reinforcing the importance of well-established genetic principles.26
Genome-wide association studies (GWAS), which were designed to discover common genetic variation that predisposes individuals to complex genetic disorders, have identified many intriguing candidate genes. However, many of the ASD genes discovered through GWAS have not achieved statistical significance in follow-up experiments.15 Some scientists have interpreted these shortcomings as failures, but this is too narrow a view given the scale of the problem. Consider that if a single gene contributes only a small amount to the risk for an ASD, then its effect will necessarily be difficult to detect (and will require a large number of patients and controls). Contrast this with Huntington’s disease, a dominantly inherited disorder where a rare but characteristic mutation in one gene (HTT) can be found in every patient and is sufficient to cause this fatal disease. [Gene mutations of this type are considered to be highly “penetrant,” which means that a high proportion of individuals with pathogenic mutations in a given gene will express the trait/disorder. By contrast, low penetrance means that only a small proportion of patients carrying a mutation in a particular gene will express the disorder associated with that gene.]
If many genes carry common mutations (some inherited and some de novo) and make varying contributions to ASDs, some larger and some smaller, then the variability of symptoms begins to look quite reasonable. Although it’s true that this complicates diagnosis, the rare candidate genes identified by GWAS are already pointing toward interesting and plausible biological pathways.27
- An important but underappreciated point: It may take only a single mutation to destroy a complex behavior, such as speech, but it takes many genes to shape the development of that trait. Like the car in the introduction, a single bad wire can cause a breakdown, but more than one wire is needed to make a car run.
Establishing causation: Where the rubber meets the road
So how, exactly, do mutations lead to the symptoms of ASDs? This is one of the most active and interesting areas of ASD research. Mutations in DNA (SNPs or structural) or epigenetic changes to genes can alter the way genes function. Often this is because those changes can result in alterations to proteins—the cell’s workhorse molecules—that are encoded by the affected genes. In some cases, the amount of a critical protein is altered, or the timing of its production. In other cases, mutations may lead to the complete elimination of an important protein or to changes in the normal patterns of when and where genes are turned on or off. (It is important to remember that not all mutations are harmful. Most, in fact, are harmless variations that have no effect on health.) Proteins make up key structural components of cells and perform essential tasks, such as communication, metabolism, catalysis of chemical reactions, and regulation of genes themselves.
The leading hypothesis for ASD causation involves neurobiology: Mutations in key genes disrupt the proper development and/or function of cells in the brain called neurons. * The CNTNAP2 gene on chromosome 7 encodes a protein that is normally part of the membranes of brain cells and is involved in signaling, but mutations can cause the protein to get stuck in other compartments within the cell, thus losing its ability to help with neuronal communication.27
Mice that have had this same gene “knocked out” (i.e., completely deleted) have abnormalities in the neural network of the brain, and the mice display symptoms much like those seen in ASDs, such as repetitive behaviors and impaired social interaction.28 Studies like this one help establish the causative links that support the correlations found through GWAS studies.
It is now possible to study the expression (when and where genes are turned on and off) of individual genes in the human brain, including ASD-linked genes.29 This opens the door to deeper investigations of pathways involved in ASDs and the effects of multiple mutations acting in concert.
Other hypotheses have been proposed for autism and ASD, including some that, if correct, might yield to relatively straightforward treatments or even prevention. For example, deficiencies in branched-chain amino acids caused by mutations in the BCKDK gene have been found in patients with autism and intellectual disability.30 Similarly, Celestino-Soper and colleagues (2012) have reported a mutation in some autism patients in the TMLHE gene on the X chromosome, which encodes an enzyme crucial to carnitine biosynthesis.31 Might carnitine and branched-chain amino acid deficiencies play a role in some autism cases? More research will need to be done, but the possibility that metabolic errors may contribute is intriguing.32
Autism and ASD serve as powerful examples of what may become a familiar story underlying the genetics of complex disease: Genetic heterogeneity will suggest important biological pathways that reveal the biochemical and molecular processes leading to disease. Those processes, and the proteins and other substances that control them, will become the targets for therapy, even as risk prediction for individuals remains challenging. This is not to imply that magic bullets will be easy to find. Indeed, researchers have identified the genetic causes of thousands of single-gene disorders, and yet cures have remained elusive. Even so, genomics is helping to reveal new sources of autism/ASD causation and will increasingly shed light on treatment possibilities.
© 2013, American Institute of Biological Sciences. Educators have permission to reprint articles for classroom use; other users, please contact firstname.lastname@example.org for reprint permission. See reprint policy.