The CNVs described in our subjects directly interrupt contactin 4 (CNTN4
), an axon associated cell adhesion molecule (AxCAMs) highly expressed in the brain, particularly in the cerebellum, thalamus, amygdala, and cerebral cortex.15
Complex interactions between cell adhesion molecules (CAMs) are of critical importance during neurogenesis and the precise functioning of neural networks. AxCAMs are believed to play crucial roles in axonal elongation along specific pathways, fasciculation of specific axonal populations, and the formation, maintenance, and plasticity of some synaptic connections.16
The expression profile of CNTN4
in human tissues indicates that the protein may have an important role in both the early growth of developing axons and in the maintenance of the adult nervous system. 16
In 2004, Fernandez et al
reported a de novo balanced translocation, disrupting CNTN4
, in a patient with the cardinal features of 3p deletion syndrome, including developmental delay and typical dysmorphic features.15
Other groups have also suggested that loss of a single functional copy of CNTN4
contributes to the developmental delay characteristic of 3p deletion syndrome.17 18
The syndrome is clinically recognised by a combination of features including growth and mental retardation, microcephaly, hypertonia, digital anomalies and dysmorphic facial features including a triangular shaped face, ptosis, hypertelorism, broad nasal root, long philtrum, down turned mouth, micrognathia and dysplastic ears.15 19 20
In this report, none of the subjects with CNVs interrupting CNTN4
demonstrated the classical 3p deletion syndrome phenotype. Quite notably, growth retardation, microcephaly, digital anomalies, hypertonia and the characteristic facial gestalt were absent and only minor, non-specific dysmorphic features were identified. This is in contrast to previous reports of CNTN4
deletion or interruption by translocation where aspects of the 3p deletion syndrome phenotype were described.15
As previously reported deletions involving CNTN4
encompassed neighbouring genes, and as translocations may similarly result in position effects on neighbouring genes, it is intriguing to observe that the intragenic CNTN4
CNVs identified in this report have been ascertained through the presence of a relatively isolated neurocognitive phenotype.15
Interestingly, mutations of contactin associated protein-like 2 (CNTNAP2
) have also been linked to ASD and/or features of the syndrome, including seizures, language regression, and mental retardation.21–24
Therefore, given CNTN4’s vital role in both the development and maintenance of the nervous system, its implication in 3p deletion syndrome, and the correlation of associated proteins with autism, we believe mutations affecting the protein’s function may contribute to ASD pathogenesis.
In our subjects, the CNVs interrupting CNTN4
were all inherited from fathers without a history of ASD; one consideration is that these CNVs are polymorphic and not pathologic. There have been a few rare reports of CNVs affecting CNTN4
in normal individuals (Database of Genomic Variants, http://projects.tcag.ca/variation/
However, most of the described variations were detected with the Affymetrix 500K EA SNP Mapping Arrays, and, for the most part, have not been validated. It may be difficult to equate data from different platforms when it comes to copy number. In addition, the normal variation reported in the Database of Genomic Variants needs to be interpreted with care. The database lists several CNVs that would be expected to cause well known syndromes, including velo-cardio-facial syndrome, 22q13 deletion syndrome, and Sotos syndrome. It is also noteworthy that an ongoing study, in our own group, utilising the same BAC array platform and analysis methods applied with this cohort, failed to identify CNVs involving CNTN4
in 560 National Institutes of Mental Health (NIMH) unrelated normal controls. Studies of other disorders (unrelated to ASD), also in our own group, have also failed to detect CNVs in this gene in 252 individuals (data not shown). If mutations of CNTN4
are incompletely penetrant, disruption of the gene, although rare, may not result in ASD in all detected cases. The significance of an incompletely penetrant mutation is perhaps best examined between families instead of individuals. This study recruited 92 subjects from 81 different families. A CNV disrupting CNTN4
is present in two families (~2.5%). However, as mentioned above, BAC microarray analysis did not detect a variation affecting CNTN4
in normal controls (data not shown). Assuming each normal individual is representative of a different family, statistical analysis of these findings with a Fisher Exact Test (http://www.physics.csbsju.edu/stats/fisher.form.html
) indicates loss of one functional copy of CNTN4
is a significant contributing factor to the development of ASD (p
0.016). Although these results need further confirmation in larger cohorts, they strongly suggest that mutations affecting CNTN4
function can cause ASD, despite their detection in a small number of reportedly normal individuals. Notably, incomplete penetrance has been described in other ASD associated mutations, including a chromosome 16p11.2 CNV that reportedly accounts for approximately 1% of all cases of the syndrome.26
The CNV was a de novo event in most subjects, but additionally was inherited in some individuals from an unaffected parent. It was also found to occur rarely in normal controls as well.
It is noteworthy that studies of large scale CNV in the human genome, to date, have not examined their cohort’s family history. Given the prevalence of ASD (1 in 150 American children), it is possible that “normal” individuals with CNVs affecting CNTN4
come from families with the disorder. Had the two fathers described in our study been part of a normal cohort, it is unlikely that the presence of ASD in their children would have been noted. Consider the case of thrombocytopenia-absent radius (TAR) syndrome. A recent study of the disease identified a 200 kb deletion on chromosome 1 in all 30 TAR patients examined.27
In a majority of cases, the deletion was inherited from an unaffected parent but the variation was completely absent from a group of 700 normal individuals. This suggests that the deletion contributes to TAR syndrome but the phenotype only develops in the presence of an as-yet-unknown modifier. Disruption of CNTN4
may affect the development of ASD in a similar manner. Imprinting, environmental interactions, or other factors may determine how mutations in CNTN4
- A genetic study of autism spectrum disorder (ASD) identified paternally inherited copy number variations (CNVs) of chr3p26 in three individuals with the disorder.
- The CNVs affected one gene directly, contactin 4 (CNTN4). CNTN4 participates in neurogenesis and functioning of neural networks. Disruption of this gene is known to cause developmental delay and mental retardation.
- Molecular characterisation of the CNVs revealed that they resulted from Alu Y mediated unequal recombination.
It is worth mentioning that subject 2A has ASD but does not carry the chromosome 3 CNV. However, he is the only child in his family to have an early regressive course, suggesting perhaps that his disease is somehow different from his siblings. Also of interest is the fact that this phenomenon has been described in ASD before. The Autism Research Consortium observed a 22q11.2 duplication in two multiplex families. In one, it was inherited from a parent, while in the other it was de novo.28
In both families, only one child diagnosed with ASD carried the duplication. Alarcon et al
describe a large intronic deletion in a multiplex family inherited in one autistic sibling but not the other.22
Given the frequency of ASD in the general population, it is possible that, on occasion, individuals from the same sibship may have the syndrome for different reasons.
Both of the CNVs interrupting CNTN4
resulted from Alu Y mediated unequal recombination. Rearrangements involving Alus, in general, are most likely to occur between repeat elements on the same chromosome located within 5 Mb of each other with high sequence identity.29
The Alu Y mediated rearrangements reported here fit these criteria, as do 43% of the Alu Y elements around CNTN4
. The high degree of Alu Y homology in this region perhaps predisposes it to Alu Y mediated unequal recombination. However, although Alu density may contribute to these recombination events, other factors likely influence the rearrangements. Analysis of other Alu-rich genes has found there is not a direct correlation between Alu density and recombination rate.30
Some other factors likely contribute to the recombination events observed in our subjects and the high Alu Y density in the region likely aids in the process.30
Our work implicates CNTN4 as a candidate gene in ASD. Ongoing efforts are underway to sequence the gene in large numbers of subjects with ASD and normal controls to identify subtle mutations that might be involved in pathogenesis.