ARE ANY SCIENTISTS TESTING THE SPERM AND SPERMATOGONIA OF THE FATHERS OF CHILDREN WITH DE NOVO AUTISM--IE FATHER 6 OK CHILDREN 7TH IS AUTISTIC?

"De novo point mutations in such genes could explain the advanced paternal age association that has been reported for autism13. There is no evidence, however, that the risk of a de novo CNV is related to the age of either parent."







Nature Medicine - 13, 534 - 536 (2007)
doi:10.1038/nm0507-534
Autism: highly heritable but not inherited
Arthur L Beaudet

The author is in the Department of Molecular and Human Genetics, Baylor College of Medicine, One Baylor Plaza, Houston, Texas 77030, USA. abeaudet@bcm.tmc.edu





The genetic basis of autism is beginning to come to light. De novo mutations in gene copy number may have a big role.


Until recently, only 5–10% of autism cases were traceable to an underlying genetic cause. Two studies now change this. Jacquemont et al.1 and Sebat et al.2 suggest that this number is actually 10–20%, and it may grow to 30–40% with further research. The advance is based primarily on the use of a recently developed high-resolution genome analysis technique to identify de novo genomic deletions and duplications of tens to thousands of kilobases (kb).

Autism is a disorder of communication and behavior that usually includes language impairment, cognitive dysfunction, limitations in social skills and, often, repetitive behaviors. The phenotype includes classical autism disorder, Asperger syndrome characterized by higher cognitive function, and pervasive developmental disorder-not otherwise specified (PDD-NOS). Many studies in North America and Europe report an incidence of about 1 in 150 children3, and the male:female ratio is typically approximately 4:1. Autism is thought to be highly heritable, largely because of a very high concordance in monozygous twins—although the concordance in dizygous twins is low, and most cases occur as isolated individuals in a family.

Autism occurs in children with sex chromosome abnormalities and with chromosomal deletion and duplication syndromes that affect every autosomal chromosome4. Individuals with autism are often divided into two groups: a syndromic, or complex, group in whom autism is accompanied by malformations or dysmorphic features (dysmorphic refers to an abnormal morphology, frequently an unusual facial appearance, as shown in Jacquemont et al.1) and an essential group who have a normal appearance. In addition, autism can be seen in individuals with single-gene disorders such as tuberous sclerosis and fragile X mental retardation. More recently, mutations in MECP2, PTEN, SHANK3 and NLGN4X have been reported to cause a small percentage of autism cases. Some of these genes regulate chromatin structure and gene expression (MECP2). Others are important in synaptic function (SHANK3, NLGN4X), which probably is abnormal in autism5.

The two studies1, 2 greatly strengthen the growing awareness that a substantial fraction of autism is caused by genomic rearrangements. Particularly in the case of deletions, loss-of-function point mutations in genes in these regions are likely to be important as well. Most of these are de novo mutations not present in the parents. The children with identifiable genetic abnormalities are often in the syndromic group, and the sex ratio for these individuals tends to be 1:1, except for X-linked disorders6, 7.

The new reports1, 2 come at a time when there is a new appreciation for human genomic variation, which can entail deletions and duplications of tens to thousands of kilobases. The extent of such copy number variation (CNV) is enormous, encompassing at least 12% of the genome and hundreds of genes8, 9. During nonallelic homologous recombination at meiosis, low copy repeats in the genome can mediate these deletions and duplications10. CNVs are extremely common under normal circumstances, but also are an important aspect of what are now frequently called genomic disorders, which are diseases caused by an alteration of the genome causing complete loss of copy, gain of copy or disruption of a dosage-sensitive gene10.

To identify CNVs associated with autism, Jacquemont et al. used a technique called array comparative genomic hybridization (array CGH)1. In array CGH, a normal control DNA and a patient DNA are labeled with different colored fluorescent dyes and hybridized to cloned DNA fragments or oligonucleotides on a glass slide. The relative intensity of the two dyes identifies differences in genomic copy number between the two samples. The investigators used an array of 3,500 large insert (80–240 kb) clones covering the genome at 1 Mb intervals. They detected eight presumed causative abnormalities in 29 autism patients (28%) regarded as 'syndromic,' meaning that their autism was accompanied by malformations or dysmorphic features. Seven of the eight presumed pathogenic abnormalities were de novo; in the remaining individual, there was an X chromosome duplication inherited from the healthy mother who was a 'carrier' for the duplication. Jacquemont et al. concluded that array CGH should be useful for genetically analyzing individuals when autism is accompanied by malformations or dysmorphic features1.

Sebat et al. performed array CGH using 85,000 oligonucleotide probes, and found de novo deletions or duplications in 12 out of 118 (10%) families with a single affected child and in 2 out of 196 (1.0%) controls2. The de novo aspect is crucial here, since we all have many inherited CNVs. Sebat et al. found five de novo events that each involved a single gene, thus focusing attention on these genes as candidates for autism2. They found that approximately 300 genes fell within the other deletions and duplications, providing a pool of additional candidate genes. Since de novo variants occurred in only 1% of the control population, it is highly likely that most of the de novo variants in individuals with autism are the primary cause of their autism. Sebat et al. emphasize that the deletions and duplications may point to genes in which point mutations may lead to autism2. Recently, this approach was successfully applied to identify autism-causing mutations in SHANK3 (ref. 11).

All of these genetic causes of autism have extremely high heritability, meaning that the mutation causes the autism, but the mutations are not inherited. This is analogous to trisomy 21, which causes Down syndrome—the heritability is very high (that is, the trisomy causes the Down syndrome) but the trisomy genotype is not inherited from a parent, but is de novo in the child. All of the genetic abnormalities discussed here would be expected to be associated with virtually 100% concordance in monozygotic twins and, for the de novo majority, very low concordance in dizygotic twins, again as is seen with Down syndrome. Thus, for a fraction of cases, this new understanding of the genetic etiology of autism fits perfectly with the previously puzzling data from twins.

The Jacquemont et al.1 and Sebat et al.2 reports focus on highly penetrant de novo deletions and duplications, but much less penetrant CNVs may also be relevant to autism. Another recent report combined a low density analysis for copy number variation with linkage studies using single nucleotide polymorphisms to search for autism loci12. This approach has the potential to identify CNVs that have smaller effects but also contribute to autism.

Many of the lesions found in the Jacquemont et al.1 and Sebat et al.2 studies had not been previously reported, suggesting that not all regions of abnormality have been detected as yet. I would estimate that only half of the de novo mutations have been detected at present (Fig. 1). Higher density arrays focused at the level of single exons are likely to be very productive in identifying smaller abnormalities that may be immediately traceable to specific genes. It is clear that both CNVs and point mutations that often affect the same genes are making major contributions to the etiology of autism. Technically, it is easier to screen for CNVs than to sequence the entire genome to discover point mutations. De novo point mutations in such genes could explain the advanced paternal age association that has been reported for autism13. There is no evidence, however, that the risk of a de novo CNV is related to the age of either parent.


Figure 1. The causes of autism according to genetic contribution and sex, with females in pink sections and males in blue.





Could de novo CNVs and point mutations explain the whole of autism? Probably not, because one would expect more frequent parent-to-child transmission than is observed for mutations of strong effect, since many individuals with autism reproduce, especially the more mildly affected Asperger patients. In addition, the male predominance would remain unexplained, because these mutations generally affect both sexes about equally, and the 'unknown' residual group of patients is predominantly male (Fig. 1).

The patients without identifiable genetic lesions tend to have a normal physiognomy and, on average, higher cognitive function, and to be even more predominantly male—with a 7–8:1 male:female ratio. Our view is that epigenetic abnormalities of chromatin that are not associated with nucleotide sequence changes might contribute to the etiology in this group; particularly given the male predominance, epigenetic abnormalities affecting the X or Y chromosome might be hypothesized. We have proposed a mixed epigenetic and genetic and mixed de novo and inherited model for autism, in which individual patients could have a genetic (mutation) or epigenetic (epimutation) etiology and these components could be inherited in some cases and de novo in others14. The de novo aspect is proving to be important, but an epigenetic component, if one exists, remains elusive. Dosage-sensitive genes—the very genes conveying phenotypes when duplicated or deleted—might be particularly susceptible to epigenetic defects. These defects might explain the relative lack of parent-to-child transmission, since such errors might be erased and reset as part of germline transmission. It is noteworthy that neither de novo events nor epigenetic events can be detected by extensive genome-wide studies currently underway to find genetic transmission or genetic linkage effects using trios of parents and children or affected pairs of siblings.