NEW YORK (GenomeWeb News) – Targets of the FOXP2 transcription factor appear to be under strong, positive selective pressure in European populations, researchers led by the Wellcome Trust Sanger Institute's Chris Tyler-Smith reported in an online, early American Journal of Human Genetics article yesterday.
Mutations in FOXP2 have been linked to speech and language disorders in humans, and though it is a highly conserved protein in all vertebrates, two amino acid substitutions have become fixed in the human FOXP2 since their split from chimpanzees. FOXP2 has been under known selective pressure, and its targets, the researchers hypothesized, may have been under recent positive selection in humans as well.
Tyler-Smith and his colleagues developed an algorithm to evaluate gene lists for evidence of positive selection, which they then applied to three sets of genes thought to be regulated by FOXP2. "Our method worked well, and overall, there was strong evidence for selection of FOXP2-regulated genes in the Europeans, but not in the Asian or African populations," Sanger's Qasim Ayub, the first author of the study, said in a statement.
Many of the genes that they found to be under selection appear to have roles in neuronal cell development, cell signaling, and immune cell trafficking, the researchers reported.
To search for genes under positive selection, the researchers developed a pipeline into which they could feed sets of genes. For each targeted gene, their algorithm came up with a list of 1,000 controls from the Ensembl database that were matched to the target gene for length, GC content, and recombination rate, which was estimated using HapMap data. Additionally, it generated lists of unique matched controls as well as unmatched controls.
The algorithm then pulled together summary statistics for positive selection from the 1,000 Genomes Project — including data from individuals of European, Han Chinese, Japanese, and Yoruban ancestry — that then were used to compare the target genes to the matched and unmatched controls. Then it identified outliers from the gene list as candidate sources for the positive selection signals.
The researchers tested their pipeline using a set of genes known to be under positive selection and a set of genes that have never been linked to being under positive selection. In the positive control set, evidence of selection was seen in European and Asian populations, though no evidence of positive selection was detected in African populations, which the researchers noted was likely due to more studies of positive selection having been conducted in populations outside of Africa. No evidence of positive selection was seen in the negative controls.
Tyler-Smith and his colleagues then applied this algorithm to three sets of putative FOXP2 targets that had been previously identified through chromatin-immunoprecipitation studies in a human neuroblastoma cell line, in embryonic mouse brain tissue, and in human fetal brain tissue. Each study generated nearly 200 FOXP2 targets, and while there was overlap between the targets captured by each screen, no target was found in all three screens.
In two of these three sets, the researchers found evidence of positive selection in the European population though not in the East Asian or African populations. In the data for the European samples, "we were able to consistently pick up significant differences between controls and human fetal brain FOXP2 targets by using a stringent Bonferroni correction cutoff," the researchers wrote.
From this, the researchers found 139 outliers with strong signals of selection in Europeans, some of which were shared across the neuroblastoma cell line, embryonic mouse brain tissue, or human fetal brain tissue. One that stood out in each of those sources was CNTNAP2, a FOXP2 target that has previously been linked to language development and impairment. But because the "selection signal overlies a substantial portion of this large gene [it] makes it difficult to assess the critical functional variant responsible for the selection signal," the investigators pointed out.
Other signals of note were traced to PON2 and RBFOX1. Both PON2, which encodes a glycoprotein, and CNTNAP2 have been linked to a pathway related to cardiac arteriopathy. RBFOX1, meanwhile, has been associated with autism spectrum disorders and is a regulator of tissue-specific splicing.
"Strong signals of selection were observed for CNTNAP2 and RBFOX1, key neurally expressed genes that have been consistently identified as direct FOXP2 targets in multiple studies and that have themselves been associated with neurodevelopmental disorders involving language dysfunction," the researchers added.
They traced other signals to genes involved in apoptosis, bone formation, and immunity.
The researchers also noted that a ChIP-seq approach could offer a "more robust set of FOX2 targets for re-evaluation," and they said that their approach "can be applied to any chosen set of genes or protein-interaction networks without any a priori hypothesis about the cause of selection."
Tyler-Smith said in a statement that they have already begun to apply this approach to other gene sets. "Our method is opening new doors to understanding how modern humans have genetically adapted to their local environments and finding candidate genes to study biological function," he said. "This approach is a practical and successful way to screen for positive selection and adaptation signals in different gene sets and populations using whole-genome sequencing data."