Most people cannot drink milk as adults without the symptoms of lactose intolerance, and most lactose intolerance is due to absence of the lactase enzyme in the gut. This presence/absence is a genetic polymorphism commonly called lactase persistence/nonpersistence, depending on whether or not lactase activity persists from childhood into adulthood.1 In Northern Europe, lactase persistence is common and many people not only drink milk, but culturally it is seen as a healthy and nutritious food. How this happened is now becoming clearer.

Lactase nonpersistence is the ancestral state, and lactase persistence only became advantageous after the invention of agriculture, when milk from domesticated animals became available for adults to drink. As expected, lactase persistence is strongly correlated with the dairying history of the population. This genetic ability to digest milk has been regarded as a classic example of gene-culture co-evolution, where the culture of dairying creates a strong selective advantage to those who can drink milk as adults, for only they can nutritionally benefit from the milk. A recent paper confirmed this link by analysing the diversity in bovine milk protein genes and showing that the highest gene diversity (and by implication the largest historical population size) is in cows from areas of the world where dairy farming is practised and the people are lactose tolerant.2 In humans, epidemiological analysis has shown that the cultural development of dairying preceded selection for lactase persistence.3 Since dairying is thought to have originated around 10 000 years ago, the selective pressure has been only for the past 400 generations. Despite this short time, there is suggestive evidence of recent positive selection: lactase persistence is associated with one haplotype, which is very common only in northern Europeans, and is distant from the ancestral haplotype.4, 5 Discovery of the possible molecular basis of this polymorphism – a single nucleotide change 14 kb away from the gene, has allowed further analysis of genetic variation associated with lactase persistence/nonpersistence.6, 7, 8

Proving that the lactase gene has been under recent positive selection in Northern Europe is difficult. As it is a recent regulatory change, codon-based methods that examine the different substitution patterns across a gene are not suitable. Instead, methods relying on allele frequency must be used – which are vulnerable to the fact that frequency patterns produced by selection can also be produced by demographic processes such as changes in population size and genetic drift. A statistic called ‘relative extended haplotype homozygosity’ (REHH) has been developed, which relies on the fact that a selected haplotype (ie a haplotype on which a relatively recent beneficial mutation has occurred and has risen to high frequency) will have an extended range of linkage disequilibrium (LD) compared with other haplotypes in the population.9 This is because the selected haplotype is young, and hence there has not been enough time for recombination to break it down. We infer that this young haplotype has been driven to a high frequency by positive selection. It is not an ideal method: since it relies on the length of linkage disequilibrium on one haplotype in relation to the frequency of that haplotype, it may be vulnerable to different sampling strategies that could alter the apparent frequency of that haplotype.10 Allele-specific recombination rates could also produce a similar effect. Nevertheless, since it compares variation on different haplotypes across the same region, it is less vulnerable to demographic changes than other population genetic measures.

REHH was used by Joel Hirschhorn's lab to provide further support for positive selection in Northern Europeans.8 It confirms that the haplotype carrying lactase persistence is almost identical for nearly 1 Mb, is therefore young and must have been positively selected to reach the observed frequency of 77% in Northern Europeans. Analysis of markers across this region showed very high genetic differentiation between European Americans (dairying) and Asian/African American (nondairying), suggesting that these markers had hitchhiked on the haplotype carrying lactase persistence. By considering the Asian Americans and African Americans to have a diversity representative of a pre-dairying ‘European’ population, a selection coefficient of 1.4–15% was calculated – consistent with the 5% previously predicted using a gene-culture co-evolutionary model.11 Did early farmers, who practised mixed farming, really rely on milk so much? There is now genetic evidence that they did, although it is still not clear why milk was so important (for discussion, see Hollox and Swallow12).

Most studies for practical reasons have focused on lactase persistence in Europe, but lactase persistence is also common in certain tribes in Africa that have a history of dairying. Is lactase persistence in these people caused by the same mutation – as would seem likely – and has it been under positive selection as well? The first part of this question has been answered by Mulcare et al.13 Their paper shows that the putative causative allele 14 kb upstream from the lactase gene is not at frequencies high enough for it to be the causative allele in Africa, even when the inherent errors in lactose tolerance testing are taken into account. There could be two reasons for this – either the allele is not causative at all and is merely strongly associated with the causative allele, or in Africans lactase persistence is due to another mutation. The first reason is possible, especially given the high LD across the region – many polymorphisms within this region will be strongly associated with lactase persistence just by virtue of being on the same huge haplotype. But functional studies from two groups show that the putative causative allele is a gain-of-function mutation increasing the expression driven from the lactase promoter in reporter gene assays in a human intestinal cell line.14, 15 So what about the second reason – a different causative mutation in Africans? Intuitively, this seems unlikely, but given the powerful selective advantage of being lactase persistent any mutation is very unlikely to be lost by genetic drift. It is possible that another mutation in the same regulatory element, a different element, or even in a trans-acting transcription factor may be responsible for lactase persistence in Africans. The answer will only be found by further genetic analysis of this locus in Africans.

As well as examining the role of this polymorphism in human evolution, this work provides an interesting case study for those concerned with finding alleles that confer susceptibility to common disease. In this case, we have a clear clinical phenotype (lactose tolerance) with a very strong well-defined Mendelian genetic component (lactase persistence/nonpersistence polymorphism), and a well-defined ‘candidate’ gene (LCT, lactase). Despite these factors, the causative polymorphism has proved difficult to discover, and the most likely causative polymorphism is located 14 kb away in an L2 repeat within an intron of another gene. Added to this, if this polymorphism is causative, then it is not the causative polymorphism in all populations. If there is a lesson to be learned from this, it is that the genetics of complex disease are likely to be very complex indeed▪