Pleiotropy occurs when a single gene affects more than one distinct and seemingly unrelated trait. Antagonistic pleiotropy occurs when one of those traits is harmful. It is widely considered to be an important foundation for the evolution of aging, in that natural selection operates strongly during early life, a period characterized by tooth and claw battles for survival and reproductive success. Evolution will select for genes, mechanisms, and biological systems that operate well early and run down later, or otherwise cause harm in later life. The adaptive immune system is an example of the type, a system that works very well right out of the gate in youth, but cannot possibly function indefinitely. It devotes resources to all pathogens encountered, and eventually simply runs out of capacity. The decline of the immune system is much more complex than that simple sketch, of course, and has numerous distinct causes, but the example serves.
In the broader sense, why doesn't the body repair itself indefinitely? The antagonistic pleiotropy hypothesis suggests that the fierce selection pressure in early life will strip away anything that isn't absolutely vital to immediate survival and reproductive success. Long-term investment in repair and maintenance simply cannot survive this evolutionary arms race, in which even a tiny loss of advantage may well lead to extinction of the lineage. This might lead us to wonder how the lowly hydra manages to be functionally immortal, actually ageless - but it is only one among countless species that all undergo degenerative aging. Perhaps we are seeing the hydra shortly before its inevitable extinction at the hands of a slightly more efficient rival.
The two commentaries here follow on from a recent paper that discussed antagonistic pleiotropy and the evidence for it. Everyone involved in the exchange appears to support the antagonistic pleiotropy hypothesis; the debate is over whether or not specific named genes in humans are clearly pleiotropic in this way, and whether the evidence in support of that position is robust. As the authors of the original paper note, the challenge inherent in human data is that it produces correlations rather than the definitive causation that can be obtained from a well-designed animal study.
Austad and Hoffmann reviewed the current state-of-the-art on what support there is for the theory of antagonistic pleiotropy and what implications this has for modern medicine regarding improving human health and longevity. Although the authors focus on examples in both wild populations and laboratory conditions, the review states that there are no compelling examples in humans where the underlying genes or alleles that carry this tradeoff have been identified. This fails to acknowledge recent studies, mostly published the last two years, where excellent progress has been made in identifying such genes and below, we describe several examples.
Two studies in 2017 uncovered evidence for antagonistic pleiotropy in genes related to coronary heart disease (CAD) and fitness, and diseases related to ageing. The first found that CAD genes in humans are significantly enriched for fitness (increased lifetime reproductive success) relative to the rest of the genome, with evidence that the direction of their effects on CAD and fitness are antagonistic. This study provides a possible reason why genes carrying health risks have persisted in human populations. The second found evidence for multiple variants in genes related to ageing that exhibited antagonistic pleiotropic effects. They found higher risk allele frequencies with large effect sizes for late-onset diseases (relative to early-onset diseases) and an excess of variants with antagonistic effects expressed through early and late life diseases.
There also exists other recent tangible evidence of antagonistic pleiotropy in specific human genes. The SPATA31 gene has been found under strong positive genomic selection. Long-lived individuals carry fewer SPATA31 copy numbers. On the other hand, its overexpression in fibroblast cells leads to premature senescence, this being the case in people having multiple copies of the gene. During human evolution, more copies of this gene have likely been favored since this protein is important in sensing and repairing UV-induced DNA damage. Unfortunately, the cost is cell senescence and premature aging.
Byars and Voskarides, responding to our review of empirical support for the antagonistic pleiotropy theory of the evolution of aging, feel that we have 'failed to acknowledge' recent human studies supporting the theory. Indeed, we mentioned no human studies because we had intended our review to present only the strongest evidence supporting the theory which has been done almost entirely in laboratory model organisms. For this reason, while we mentioned a few studies from natural populations, we emphasized how such nonexperimental studies could be consistent with the antagonistic pleiotropy mechanisms, but could not be cleanly attributed to it. Experimental studies establish cause-and-effect in a way that correlational studies cannot.
It is an unfortunate truth about research on humans that because experimental studies are often impossible, results are almost inevitably correlational, which in our view makes virtually any single study highly suggestive at best, but never compelling. To illustrate why, we consider one of the studies adduced by Byars and Voskarides, although we could have chosen any of the others. That study identifies numerous human alleles pre-disposing individuals to coronary artery disease (CAD) but also conferring reproductive advantages early in life. This is a very nice study given the limitations of human research. The best that could be done with available data was done. Note, however, that one of the first lessons of statistical reasoning is that correlation does not equal causation and, yes, genomic associations are correlations.
Our point in noting these things is certainly not to denigrate the study by Byars et al. or the other studies cited. These are some very fine studies using state-of-the-art genomic analyses. We simply wanted to explain why we consider such studies less compelling as support for the antagonistic pleiotropy theory than experimental studies done in model laboratory organisms with specific and purposeful manipulation of specific individual genes.