Local protein synthesis in neurons

are likely to be harmful and prevented from spreading in the population by natural selection. This implies no functional importance for an Alu sequence itself, but merely that, as the deletions of Alus are very unlikely to be precise, a deletion event removing an Alu is also likely to remove valuable sequences around it, and the chromosome bearing the deletion will be lost by selection. The explanation favoured by the authors for Alu enrichment in generich regions is that of positive selection in favour of Alus in GC-rich DNA. This theory, however, cannot explain the observations. The data show that Alu sequences up to five million years old are not enriched in GC-rich regions. But in human population genetics, estimated times to common ancestry of typical genomic regions show that Alu sequences which are five million years old have already been fixed (found in all individuals) in the population. This observation is also what would be expected from neutrality and genetic drift, given the human effective population size. (Alu sequences which are truly advantageous will spread to fixation much more quickly.) Earlier human ancestors would also be expected to have had similar fixation times for Alu insertions. Yet it is only during the spread to fixation of Alu sequences that positive natural selection has any opportunity to act. Thus, an increasing abundance of Alu sequences in GC-rich DNA as they age beyond five million years cannot be the result of natural selection for positive functions of Alu insertions.