|  Autism is a
neuropsychiatric disorder that exhibits high heritability and is considered
to have a
complex genetic etiology. A sibling of a child
with autism has a 25 to 50 times greater risk for developing autism than
someone in the general population. Autism displays both clinical and genetic
heterogeneity, as reviewed in last month’s column. A different set
of genes may confer risk in different families or individuals (genetic
heterogeneity), and different siblings in a given family may have a different
clinical presentation (clinical heterogeneity). Research groups have been
trying to identify susceptibility genes through genome-wide linkage studies
and candidate gene analysis. The former typically identifies regions of
the genome that are more frequently shared by affected sibling pairs with
autism than would be predicted by chance. The latter typically takes the
form of allelic association studies in which alleles of a given gene are
tested for evidence of preferential transmission to affected family members
(family-based) or for differences in allele frequency between autism and
control populations (case-control). Despite evidence for high heritability,
data emerging from genomic linkage screens suggest that there may be 10
or more major genes underlying autism susceptibility and that different
sets of genes may be responsible for risk in different families.
In addition to linkage and candidate gene studies, chromosomal abnormalities
may provide important clues to the identification of genes underlying autism,
or any genetic disease. For example, several chromosomal translocations
and other abnormalities have been identified that affect the region of
7q in virtually all genomic linkage studies of autism. Traditionally in
human genetics, chromosomal abnormalities are of substantial utility in
identifying a discrete region harboring a disease gene. The most frequent
chromosomal abnormality seen in autism populations involves duplication
of sequences in a region on the proximal part of the long arm of chromosome
15, specifically the interval 15q11-q13. These duplications typically take
one of two forms: (1) tandem duplication of a 4–5 million base-pair
(Mb) region corresponding to 15q11-q13, or (2) supernumerary pseudodicentric,
inverted, and duplicated regions of chromosome 15 [so-called inv dup(15)
or idic(15) marker chromosomes] that now contain two additional copies
of a larger region. These duplications are associated with substantial
risk for autism when derived from maternal but not paternal chromosomes.
This parental-specific association suggests a genomic imprinting effect
and makes relevant consideration of two disorders that result from interstitial
deletion of the same 15q11-q13 region.
(Figure 1)
Genomic imprinting is an epigenetic phenomenon in which gene expression
occurs preferentially from one or the other parentally derived homologous
chromosomes. Imprinted gene expression can result in parent-of-origin effects,
thus producing complicated inheritance patterns for disorders involving
imprinted genes. Paternal-specific deletion of the 15q11-q13 interval results
in Prader-Willi syndrome (PWS), while maternal deletion gives rise to the
quite distinct Angelman syndrome (AS) phenotype. Although a common region
is deleted in both disorders, PWS results from a loss of expression of
multiple imprinted, paternally expressed genes, whereas AS arises from
the loss of function of a single, imprinted, maternally expressed gene
encoding a ubiquitin-protein ligase (UBE3A).
PWS presents with infantile hypotonia, failure to thrive, and feeding difficulties
due to poor suck reflex. By age 2 years, individuals with PWS begin to
develop hyperphagia and secondary obesity, easily the most manifest aspect
of the phenotype. People with PWS display mild to moderate cognitive impairment
and physical findings including decreased stature, small hands and feet,
almond-shaped eyes, and hypogonadism. Persons with PWS typically have behavioral
abnormalities, including aggression, self-abuse, preoccupation with ordering
and arranging, resistance to change in daily routines, and food foraging.
By contrast, AS is more severe, with profound mental retardation, absent
speech, epilepsy, gait ataxia, hand-flapping, and inappropriate laughter.
Other causes of these disorders are uniparental disomy (UPD; inheritance
of both copies of a given chromosome from the same parent) for chromosome
15 and imprinting defects for both disorders, and maternal-specific point
mutations of UBE3A for some patients with AS (see columns from June and
July 2001).
Numerous cases of chromosome 15 duplication associated with autism have
been described, and collectively these reports permit some conclusions.
Idic(15) is typically seen affecting maternal chromosomes, is caused by
errors in the normal segregation of chromosomes during meiosis, and is
possibly correlated with advanced maternal age. The phenotype associated
with idic(15) cases is typically more severe than that described for interstitial
duplication of 15q11-q13. Many case reports of idic(15) patients describe
autistic phenotypes with relatively profound cognitive impairment, learning
disability, developmental delay, speech impairment, seizures, poor motor
coordination, hypotonia, joint laxity, and motor stereotypies. Some of
these patients have even been described as “Angelman-like,” reflecting
the neurological symptoms and significant cognitive impairment. Interstitial
duplications have been informative in defining the maternal specificity
for autism risk. Many, but not all cases of maternal duplication meet criteria
for an autism diagnosis using standard measures derived from the Autism
Diagnostic Interview and Autism Diagnostic Observation Schedule. The most
detailed study comparing the effect of maternal versus paternal 15q duplication
showed that none of the paternal duplication cases exhibited evidence for
a pervasive developmental disorder, whereas a significant minority of the
maternal duplication patients met criteria for autism or related pervasive
developmental disorders. A family described by Cook and colleagues was
similarly informative, as two children inheriting a duplication from their
mother had autism or autism-spectrum phenotypes, while the mother, carrying
the duplication on her paternally derived 15, was clinically normal. Thus
maternally derived duplication of 15q11-q13 confers significant risk for
the development of autism.
While the idic(15) duplication spans a larger genomic interval, these cases
also have a greater copy number of 15q11-q13 gene loci. Thus increasing
copies of 15q11-q13 may be correlated with increasing severity of phenotype
due to a gene dosage effect. Maternal specificity of duplications suggests
an imprinting effect, presumably related to expression of the imprinted,
maternally expressed genes in the duplicated segment. In addition to UBE3A,
another maternally expressed gene (ATP10C) is located in this region. The
ATP10C gene product is believed to function as a phospholipid transporter
protein that may be involved in CNS signaling. Therefore, overexpression
of one or both of these genes, caused by an increase in maternal gene copy,
could represent a major underlying molecular factor in the development
of autism phenotypes associated with chromosome 15 duplications. However,
a contiguous gene duplication effect could also be important. That is,
overexpression of multiple genes in this region may contribute to the phenotype.
This would help explain the more severe phenotype in association with idic(15)
markers, given the greater number of duplicated genes. Furthermore, maternal
UPD 15 provides a theoretically analogous situation with regard to overexpression
of imprinted, maternally expressed genes; however, maternal UPD 15 differs
in that nonimprinted genes in the region are presumably expressed at normal
levels. Although autistic features have occasionally been reported in maternal
UPD 15, autism is not a common finding. Taken together, these data support
the involvement of both imprinted and nonimprinted genes in 15 duplication-derived
autism.
Chromosome 15 duplications are present in only an estimated 1% to 3% of
individuals with autism. An important question, therefore, is whether chromosome
15 genes play any role in the inherited risk for autism in the overwhelming
majority of subjects without chromosomal abnormality. Results from genomic
linkage studies with regard to chromosome 15 are mixed. Of seven linkage
screens, three have supported linkage to this region. Existing genetic
data converge on a cluster of three g-aminobutyric acid (GABAA) receptor
subunit genes (b3, a5, and g3). The GABA receptor genes make attractive
functional candidates given a developmental role for GABAergic transmission
in establishing neuronal connectivity and a central role in the maintenance
of inhibitory tone in the adult brain. Moreover, GABAA receptor agonists
treat a number of conditions related to the autistic phenotype including
seizures, anxiety, and social phobia. Candidate gene studies in this region,
particularly focused on the GABAA receptor subunit cluster, have repeatedly
identified the b3 (GABRB3) gene. Several groups have reported significant
evidence for allelic association at this gene using microsatellite markers.
One group has observed association in the maternal expression domain, at
DNA marker D15S122, located in the promoter region of UBE3A.
The identification and analysis of phenotypic subsets in studies of genetic
linkage or allelic association will likely increase power to detect genetic
effects in autism. Rather than examining chromosomal sharing in a heterogeneous
population of autism families, genetic studies may be performed using only
those families that exhibit, for example, significant deficits in language,
score high on Autism Diagnostic Interview items related to compulsions
and rigidity, or display more (or less) social impairment. This approach
is likely to identify genetically more homogeneous groups of families;
analysis of these families should reveal more significant findings for
gene locus variants that specifically affect a particular phenotype domain.
Support for this strategy is found in efforts to identify those features
of the broader autism phenotype that are significantly correlated between
siblings and in the frequent observation of these traits in family members
who do not have autism. Thus a susceptibility allele at a given gene locus
may have a somewhat more specific impact on language or rigidity, for example,
and not influence all symptom domains. Several recent reports using language
as a basis for defining who is affected reveal significantly increased
evidence for genetic linkage to regions on chromosomes 7q, 2q, and 13q.
Preliminary studies in this regard also appear to hold promise for the
detection of chromosome 15 disease-associated genes.
Phenotypic comparisons of the neurobehavioral disorders resulting from
15q11-q13 defects reveal intriguing aspects of overlap between PWS, AS,
and the autism-spectrum disorders. Patients with AS have severe cognitive
impairment so direct comparison is problematic, but motor stereotypies,
poor motor coordination, seizures, and significant language impairment
are features common to both AS and autism. Several interesting behavioral
manifestations are seen in both PWS and autism, including compulsions,
self-abuse, and comparatively unimpaired or even superior performance in
certain discrete cognitive domains. A typical pattern of intellectual disabilities
with areas of relative sparing has long been appreciated in autism and
can range from performance on par with mental-age–matched controls
(so-called splinter skills) to phenomenal savant abilities. These skills
often include visuospatial, computational, mnemonic, and musical talents.
Individuals with PWS are often regarded as gifted at solving jigsaw puzzles.
It is possible, therefore, that there is a biological and genetic basis
for the commonalities between these disorders. This remains to be demonstrated,
but it is an area of great interest to those groups that study the genetics
and psychopathology associated with defects of this chromosomal region.
To localize genetic susceptibility to autism on 15q, several groups are
pursuing an assessment of allelic association for the region from UBE3A through the GABR genes using single nucleotide polymorphism (SNP) markers.
SNPs are single base-pair changes in DNA sequence that occur on average
every 1,000 base-pairs in the genome. The vast majority of these variations
are benign changes, but some may produce physiological effects, giving
rise to normal human variation but also risk for disease. SNPs are a valuable
tool for mapping disease susceptibility, since they allow investigators
to distinguish between the two copies or alleles of a gene. Given the frequency
with which these variations occur in human DNA, they can be used to create
a dense map for examining allelic association. Inasmuch as adjacent SNPs
are rarely separated by recombination events, particular alleles at nearby
SNPs are most often transmitted together. This association of alleles at
adjacent markers is known as linkage disequilibrium (LD). Exploiting LD
relationships, investigators can detect allelic association at markers
near a functional susceptibility variant, even if those markers themselves
are not involved in disease. LD studies seek to identify an underlying
haplotype (a preserved segment of an ancestral chromosome) that contains
a susceptibility variant by detecting association at individual SNPs. An
extension of this strategy examines transmission of haplotypes directly
to determine whether a broader region may contain a susceptibility allele.
Once a region is identified in this way, the surrounding sequence can be
directly screened for functional variants that confer autism risk. In the
absence of a single ancestral susceptibility allele, where many different
risk alleles exist (allelic heterogeneity), this strategy is limited. A
complementary approach is to directly screen genes in affected individuals
to identify rare disease-associated mutations. An ever-evolving understanding
of regional haplotype structure, continual advances in rapid SNP genotyping,
and the implementation of methods such as phenotypic subsetting to facilitate
genetic analyses, bring new insights, power, and promise to the detection
and localization of a chromosome 15 autism susceptibility allele.
top of page
|
