There is a well-established difference in cholesterol levels between Americans of European and African descent. In particular African Americans generally have higher HDL levels. HDL is known as the “good” cholesterol and has in many studies been associated with protection against cardiovascular disease. Curiously, African Americans generally have higher rates of cardiovascular disease – their high HDL levels do not seem to provide them much protection. Although socio-economic and environmental factors including diet certainly contribute, these factors alone do not appear to fully explain the difference. There appears to be genetic variants that cause African Americans to have higher HDL levels but less protection against cardiovascular disease. Finding these genetic causes is an area of much active research, and some studies suggest that the reduced HDL protection in African-Americans is related to paraoxonase activity. With the increase in GWAS studies focusing on African-Americans we will likely learn more about these factors within the next few years. But there is another question buried here – an evolutionary question. Why would there be differences in HDL levels and HDL functionality in different geographic areas of the world? One possibility is that it is purely random; genetic drift has caused these differences. A perhaps unlikely explanation, but an explanation that we nonetheless want to rule out before speculating too much about the adaptive reasons for the observed differences.
A new study from our group published in Molecular Biology and Evolution with Anna Ferrer-Admetlla as lead author, sheds some light on this. Behind the technical title “On detecting incomplete soft or hard selective sweeps using haplotype structure” hides a few results that might help us better understand the evolutionary underpinnings of HDL biology in African Americans. Using a new haplotype based statistic, we show that APOL1, the gene encoding the major protein component of HDL, is one of the genes showing the strongest signature of natural selection in Yorubans – a group from which most African Americans descent. We do not see a similar pattern in any other group investigated, including Maasais from East Africa. Most other groups show evidence for selection primarily in immune and defense related genes, except for the Maasais which show most evidence for selection relating to the lactase genes – a story that has already been investigated in great detail by Sarah Tishkoff. In addition, we find several other genes relating to cholesterol metabolism which also show strong evidence of selection in Yorubans, including CD36, a gene implicated in the binding and internalization of oxidized LDL (the “bad” cholesterol). So clearly, something serious happened evolutionarily with cholesterol in Yorubans. We can probably rule out that the differences in HDL levels and functionality between African Americans and other groups are simply a consequence of genetic drift – natural selection most likely caused these differences.
This then raises the next question: which phenotypes did selection act on? Without a time machine we will never know. Even though we can identify certain phenotypic effects of the genetic variants that selection has acted on, we cannot conclude that selection worked to change these specific phenotypes. Many, if not most, genes are highly pleiotropic – they affect multiple different phenotypes. Which of these phenotypes where the primary target of selection can be difficult to discern. However, one obvious explanation for the selection acting on APOL1 and other cholesterol related genes in Yurubans, might be changes in diet. Demands on cholesterol activity may depend on diet, and there may be trade-offs relating to the efficacy of fatty acid uptake, energetic costs, and risk of cardiovascular disease that have imposed different selective regimes in different parts of the world. While this explanation seems obvious, it doesn’t really explain why selection has targeted only people in West Africa. Many other groups around the world have experienced changes in diet. Our hypothesis is instead that the selection is driven by pleiotropic effects relating to defense against parasites. APOL1 is involved in parasite killing, particularly trypanosome killing. APOL1 triggers uncontrolled osmotic swelling of the lysosome in the parasite, an effective mechanism for elimination of parasites. Similarly, CD36 harbors genetic variants associated with susceptibility to malaria. The pathogenic environment differs greatly between different geographic regions, and West Africa has certainly historically been one of the areas of the world with highest parasitic load. So one likely explanation is that the selection on genes such as APOL1 and CD36 observed in Yurubans is in response to selective pressures imposed by parasites – not by changes in diet. When African Americans today have different HDL levels and functionality than other groups, it may be a byproduct of past selection acting on their ancestors in Africa in defense against parasites.
I decided to write this rare blog entry, in part because I wanted to draw attention towards these results from the Ferrer-Admetlla paper – results that otherwise might remain buried in the technical aspects of the paper. But this example also well illustrates the challenges in evolutionary biology in identifying adaptive causes. It is often quite easy to identify the footprints of natural selection in DNA data, at least when natural selection is very strong. We have a slew of good methods for detecting natural selection. Unfortunately, we will in most cases never know for sure which phenotypes were targeted by selection. Evolutionary biology is in part a historical science – and as such we have to live with some ambiguity: we can detect the footprints of past selection – but we may never know for sure which phenotypes selection acted on. There is not a simple experiment that can determine what happened in the past.