Autism is a pervasive neurodevelopmental disorder characterized by deficits in language development and social interaction, as well as stereotypical, repetitive behaviors. The etiology of autism is largely unknown. Family and twin studies have provided compelling evidence for a strong genetic component in most idiopathic cases. Several recent candidate gene studies have suggested that alleles of WNT2 and the reelin gene (RELN), two genes involved in distinct aspects of neurodevelopment, confer greater susceptibility to autism. We screened WNT2 for DNA polymorphisms by sequencing all exons and adjacent intronic regions in 24 autistic patients, and identified not only the WNT2 variants reported previously (two common single-nucleotide polymorphisms (SNPs) in the 5' upstream region and the 3' untranslated region (UTR), respectively), but also two new SNPs in its 3' UTR. We genotyped all four WNT2 polymorphisms and a polymorphic trinucleotide repeat in the 5' UTR of RELN in 107 families with multiple autistic children, and evaluated evidence for association between these variants and autism by the transmission disequilibrium test (TDT). Our results revealed no deviation from the null hypothesis of no association. Our interpretation of these findings is that it is unlikely that DNA variations in RELN and WNT2 play a significant role in the genetic predisposition to autism.

A recent report suggested that the HoxA1 and/or HoxB1 genes play a role in susceptibility to autism. To determine whether these findings could be confirmed, we screened these genes for DNA polymorphisms by sequencing all exons in 24 individuals with autism. We identified the same sequence variants in the genes that appeared in this report, which include one single-base substitution variant in HoxA1 and a common haplotype in HoxB1. We performed an association study by applying the transmission disequilibrium test to detect possible association of these variants to autism in 110 multiplex families. Our results demonstrated no deviation from the null hypothesis of no association. We have also separately examined transmissions within individual mating types, for paternal versus maternal alleles, to affected versus unaffected children, and for transmission to affected boys versus girls. None of these subsets revealed significant deviation from the null expectation. Our interpretation of these findings is that it is unlikely that HoxA1 and HoxB1 play a significant role in the genetic predisposition to autism.

Attempts to identify regulatory sequences in the human genome have involved experimental and computational methods such as cross-species sequence comparisons and the detection of transcription factor binding-site motifs in coexpressed genes. Although these strategies provide information on which genomic regions are likely to be involved in gene regulation, they do not give information on their functions. We have developed a functional selection for promoter regions in the human genome that uses a retroviral plasmid library-based system. This approach enriches for and detects promoter function of isolated DNA fragments in an in vitro cell culture assay. By using this method, we have discovered likely promoters of known and predicted genes, as well as many other putative promoter regions based on the presence of features such as CpG islands. Comparison of sequences of 858 plasmid clones selected by this assay with the human genome draft sequence indicates that a significantly higher percentage of sequences align to the 500-bp segment upstream of the transcription start sites of known genes than would be expected from random genomic sequences. We also observed enrichment for putative promoter regions of genes predicted in at least two annotation databases and for clones overlapping with CpG islands. Functional validation of randomly selected clones enriched by this method showed that a large fraction of these putative promoters can drive the expression of a reporter gene in transient transfection experiments. This method promises to be a useful genome-wide function-based approach that can complement existing methods to look for promoters.

No abstract