SNPs Reveal Natural Selection in Human Populations

CAMBRIDGE, Mass. — Some people carry better genetic armor for resisting infectious disease than others. For example, many Africans have allelic variants of several different genes that provide some resistance to malaria.

Geneticists would like to know whether such resistance arose through selective pressure or merely represents random mutations that remain in the population. One reason for seeking an answer to the question is the practical application it could give. If natural selection within human subgroups has occurred, and if researchers had a method for reliably uncovering such selection, the strategy might pinpoint selected variants of other genes that conferred resistance to disease. Information of that sort could lead to new treatments or better methods of disease prevention.

Genome Center researchers, led by Eric Lander, have just devised such a strategy. With it, postdoctoral fellow Pardis Sabeti and colleagues found striking evidence for natural selection of allelic variants of two different genes that are associated with malaria resistance in certain Africans. The work appears in the October 9 online issue of Nature.

Sabeti’s approach involved studying haplotypes around two genes (G6PD and CD40 ligand) for which a particular allele confers resistance to malaria. Haplotypes are sets of closely linked genetic markers present on one chromosome that tend to be inherited together; in this case the researchers used single nucleotide polymorphisms (SNPs) around the variants of interest to define the haplotypes. They compared haplotype frequencies among six populations; three from African, African-Americans, European-Americans, and Asians.

The study looked for evidence that a particular haplotype associated with the resistance variant arose faster in the population than would have happened by chance. As Sabeti put it, "New mutations take a long time to rise in frequency in a population, but the mutations that protect humans from disease are enhancing the survival of its carriers and may therefore rise in frequency more rapidly." The researchers predicted that, if selection occurred, they would see a particular haplotype, stretching over a long distance of DNA, that extends much further than similarly prevalent haplotypes.

Sabeti and her colleagues verified that result, both for the G6PD and CD40 ligand genes. By contrast, inspection of haplotypes around 17 other regions of DNA, chosen at random, showed no such evidence of selection pressure.

The power of this new strategy lies in generating haplotypes with large numbers of SNP markers covering over 400,000 nucleotides in length, then creating a computer program that uses all the genetic information in the analysis. As Sabeti noted, the case-control study that first linked a particular G6PD variant allele with a decreased risk for malaria analyzed data from over 2000 people and reported a difference that was significant at a p value < 0.02. By contrast, the current study inspected DNA from just 78 individuals and obtained differences that were significant at a p value < 0.0008.

The Whitehead team showed the superiority of their new test compared to several statistical tests commonly used in population genetics; all the others lacked the power to show that the data were anything other than random. The robustness of the new test impressed biostatistician Warren Ewens (Unversity of Pennsylvania), the originator of the first such test for selection pressure, on which the more recent tests are based. "Using the large number of SNPs gives them a more informative data set" said Ewens. Most other tests use one or just a few makers whereas the new test examines a cluster of SNPs nearby the region associated with resistance within a background of more distant SNPs.

The current study examined genes for which there was proof (in the case of G6PD) or strong supportive evidence (in the case of the CD40 ligand) that a given allele reduces the risk for malaria. Given the robustness of the results, Sabeti notes that she, David Reich, and her other colleagues at the Genome Center are anxious to start looking for evidence of genetic selection for resistance to other infectious agents.

Both the G6PD and CD40 ligand genes reside on the X chromosome in humans, and the current study of them was restricted to an inspection of males only. Hence, the researchers avoided any confusion due to having two alleles per individual. However, Sabeti said that using a large number of SNPs in the analysis permits some assumptions that get around this problem. Therefore, the group has indications that the method will lend itself to studies of autosomal genes as well. Consequently, Sabeti and her colleagues hope that this new strategy will allow researchers to quickly spot potential candidate genes for their involvement in certain diseases.