The evolutionary theory of aging provided a solution to the question of why we age in the mid-20th century. Experimental work demonstrated the soundness of its principles in the latter portion of the century. The new omics technologies of the 21st century have now provided the possibility for serious interventions and possible treatments for aging. This essay acquaints the reader with the evolutionary theory and its mechanisms. It concludes by asking what the consequences might be for developing such technologies.
In Greek mythology fate was embodied by three sisters, Clotho (life), Lachesis (lifespan), and Atropos (death). As the fates would have it, on that snowy day that I chose to return to science, I would become a part of one of the most significant scientific revolutions of the 20th century concerning the responsibilities of the fates. The ancients pretty much uniformly felt that aging, disease, and death entered the world due to some curse from the gods. For the Greeks it was Pandora’s curiosity; for the Hebrews it was Satan’s tempting of Eve to eat the fruit from the tree of knowledge. The Old Testament (Hebrew Bible) recounted conflicting ideas concerning how long individuals might live, with Methuselah supposedly living over 900 years, but Psalms describing the life of a man as only seventy to eighty years (Psalms 90:10). But as poetic as these ideas are, they really leave us with no mechanistic understanding of why organisms age or even a clear definition of what aging is (1).
The application of evolutionary theory to the problem of aging is one of the greatest revolutions in the history of science. Kuhn describes “scientific revolutions” as a point at which the established paradigm can no longer stand (2). Prior to the 19th century, there was really no scientific approach to aging. At best, aging was conceived in the same sense as thermodynamic theories, such as “wear and tear.” Examples of wear and tear include the breakdown of mechanical parts over time, such as the wings of insects (3). Of course, such theories only claimed that all things would eventually break down, but provided no insight as to why the maximum life span of a mouse was a couple of years and that of trees might be hundreds to thousands.
The scientific revolution that established the evolutionary foundation for aging began with the thinking of Alfred Russell Wallace and August Weismann in the 19th century. Wallace’s ideas did not progress beyond group seIectionist notions, e.g. natural selection favoring the loss of older organisms to make way for younger, reproductive ones. Whereas Weismann recognized that physico-chemical processes could not explain the duration of life and that therefore the key to understanding aging in metazoans might be found in the separation between somatic and germ tissue. The theoretical formulations for the evolutionary theory of aging were laid by some of the greatest scientists of the 20th century, e.g. Ronald A. Fischer, H.T.J. Norton, J.B.S. Haldane, Peter Medawar, George C. Williams, William D. Hamilton, Brian Charlesworth, and Michael Rose. By 1928 Norton and Haldane had reasoned that natural selection impacted evolutionary fitness in an age-specific manner. In 1930, Fisher recognized that the declining survival of adults with age could be due to the declining “reproductive value” of older individuals (4). He defined reproductive value as the expected future contribution of an individual of a particular age to the future reproduction of a population, calculated in reference to the rate of population growth (5). The reproductive value that he calculated increased from the age of first reproduction to some maximum point, and then declined until it reached zero. This concept would play a major role in Peter Medawar’s formulation of the declining force of natural selection with the age of adult somatic tissue. This meant that an allele whose negative impacts occurred after the point at which an individual’s future reproductive value was zero could have no impact on future generations. This hypothesis was further developed by William D. Hamilton, who demonstrated that the force of natural selection weakens with adult age in organisms with a separation of their germ and somatic tissue (6).
Thus, the simplest way to understand why metazoan organisms age comes from Peter Medawar:
The force of natural selection weakens with increasing age—even in a theoretically immortal population, provided that is exposed to real hazards of mortality. If a genetical disaster…happens late enough in individual life, its consequences may be completely unimportant. (7)
Medwar’s idea set up the only population genetic mechanisms that could account for aging: antagonistic pleiotropy (AP) and mutation accumulation (MA). AP refers to genes that have multiple effects (pleiotropy) under different circumstances (antagonistic). Thus a gene that is beneficial during early life may be detrimental during late life. MA refers to mutations that have no effect of evolutionary fitness in some circumstances, but are deleterious in others, so that a gene may be neutral in early life but have deleterious effects in late life. In addition, this theory of aging suggested a simple laboratory experiment. If the force of natural selection declined with the age of the adult soma, it followed that the timing of reproduction across adulthood would profoundly influence the rate at which that force declined. Several researchers recognized that, and experiments in insects began to test this idea in the late 1960’s. Wattiaux, for example, conducted the first “postponed” aging experiments with Drosophila pseudoobscura and D. subobscura in 1968 (8, 9). These experiments suffered in detail, so the first clearly demonstrated success utilizing experimental evolution to postpone aging was Rose 1984 and Luckinbill 1984 (10, 11). P and MA operated to determine the pattern of senescence in both of these experimental systems. AP was shown to cause a tradeoff between early reproduction and delayed survivorship, as well as in the physiological systems mediating this (e.g. stress resistance, flight capacity). Thus fly stocks displaying delayed senescence were capable of reproducing later in life and displayed superior physiological performance across their life span compared to the stocks with early reproduction, shorter life, and less physiological capacity. MA was demonstrated as well, with some physiological traits showing no relationship to early life fitness but were associated with late life decline.
Subsequent studies of senescence in the Rose system demonstrated that aging was influenced by genetic variation at 100’s to 1,000’s of genetic loci. Senescence was influenced by changes at individual positions within the genetic code (called single nucleotide polymorphism, SNPs). There was also structural variation in the genetic code (including insertions or deletions of nucleotides, called “indels”) and transposable genetic elements, which code for their own replication and are capable of inserting themselves at different points in the genome (12).
Mutation Accumulation (MA) and Antagonistic Pleiotropy (AP)
The frequency of genes that have age-specific effects will be determined according to Table 1. Any allele that has a negative impact on fitness (determined as the product of age specific survival (lx) and age-specific reproduction (mx) will be removed from the population by purifying selection. This means that the action of these genes are not relevant to the pattern of aging within any species. By necessity they will be exceedingly rare, the frequency determined by the mutation rate and the selection against the gene (as those who carry them rarely if ever reproduce). An example of this in humans is the disease Hutchinson-Gilford progeria syndrome (misnamed as “premature aging”). This disorder is characterized by short stature, low body weight, early loss of hair, lipodystrophy, scleroderma, decreased joint mobility, osteolysis, and facial features that resemble aged persons. The majority of these individuals display a dominant mutation in the lamin A gene (LMNA 1q22). The frequency of this condition was 1 per 8 million newborns in the United States between 1915—1967 and 1 per 4 million births in the Netherlands from 1900—2005. The mutation is found in all world populations at different frequencies due to genetic drift. Genetic drift refers to the random fluctuation in allele frequencies in populations due to small population size.
MA refers to alleles whose impacts on fitness are neutral in early life, but can have deleterious effects in late life. The frequency of these variants will be determined by genetic drift. MA has been demonstrated in a variety of experimental organisms, such as plants, fruit flies, and round worms. In a study of the plant Arabidopsis thaliana, grown in optimal conditions, it was shown that compared to ancestor, five derived lines accumulated 99 base substitutions and 17 indels over 30 generations (13). The benign environment used to propagate the plants allowed these mutations to be neutral and thus accumulate in the populations.
There is evidence that MA influences aging in humans. A recent study examined disease progression from data derived from the UK Biobank and found that a genetic variation associated with multiple diseases (cardiovascular, diabetes, osteoporosis, cataracts, immunological, neurological, muscoskeletal, gastrointestinal, and respiratory) fit the mutation accumulation mechanism (14). The diversity of age-related diseases associated with mutation accumulation is not surprising, simply because of the sheer numbers of genetic variants that display neutrality in early life. The impact of these variants will differ by population however, as their frequency will also be determined by genetic drift. Thus a study of mutation accumulation variants associated with late life disease generated from a European population might not replicate entirely for populations from Africa or Asia. Nor would these variants if they were determined from Finns (a population whose allele frequencies were largely determined by genetic drift) replicate to the rest of Europeans.
AP is the most insidious of the population genetic mechanisms accounting for aging. This is because the beneficial character of these mutations in early life means that positive natural selection will drive them to fixation in the population (every organism in the population will have them). AP has been also demonstrated in a variety of experimental organisms. For example, whole genome sequencing of fruit flies exhibiting postponed senescence show starkly different patterns of SNP variation between early-reproduced (high early fecundity, short life) and delayed-reproduced (low early-fecundity, and long life) Drosophila melanogaster (15). Again there is evidence that this mechanism operates in humans. The study of the British Biobank data (as well as from 1000 Genomes data for other Europeans) also demonstrated the existence of SNPs whose frequency was driven by antagonistic pleiotropy. They found that loci ABCG8 and ABCG5 showed antagonistic pleiotropy with high cholesterol, lipid related diseases, as well as with gall bladder disease and cholelithiasis. The locus ADH1B was shown to be in AP with cardiovascular disease and gout; while SLC39A8 showed AP with osteoarthritis (16). Thus, this study showed that diseases that are common during human aging, are produced by the action of genes that were beneficial during early life.
A notable omission of this UK Biobank study is that it choose not to examine loci associated with cancer. For example, mutations in BRCA1/2 genes cause increased risk of ovarian and breast cancer. One study showed that women of European descent (Utah, USA) living under natural fertility conditions (born before 1930) had more children, shorter interbirth interval, later age of last birth, and higher post-reproductive mortality than women without the mutations (16). Several other cancer related loci have also been shown to increase early life fitness such as AR, TP53, PTPN11 (17).
Theory suggests that we should see a clear AP relationship between loci associated with early life fitness (development, tissue repair) and malignant neoplasms after net future expected reproduction is zero. Figure 1a shows age-associated cancer mortality in US adults from 2020, while figure 1b shows the age-associated birth rate of US females in that same year. The data show that the mortality for cancer (derived from all malignant neoplasms) does not begin to increase until after the age-associated birth rate of US females is essentially zero (net future reproduction = 0). Females are used for this comparison, as it is much easier to quantify their reproductive output compared to males. Also, while we know that males are physiologically capable of reproduction later in life than females, for the vast majority of males, their reproductive success is determined by the age of their partner. Finally, the age-specific patterns of gene expression displayed by a population is set by natural selection that occurred on their ancestors, but there is good reason to expect that the age-specific pattern of cessation of female reproduction has remained relatively stable well into the human past (18).
The pattern of cancer-related gene expression could be explained by either the MA or AP mechanisms, but in the case of cancer there is an excellent reason to suppose that the frequency of cancer predisposing alleles is driven by AP. First, most cancers occur in stem cells. These cells play a vital role in the maintenance of tissues in multicellular organisms. They replace cells in tissues such as bone marrow, as well as the epithelia of the lungs, gut, and skin. In addition in placental mammals the embryonic stem cells of the invasive placenta contain much of the genetic program needed to produce cancer. Thus, fewer mutations are required in such stems to make them metastatic. Thus, traits that are advantageous to the organism during development and adulthood (+) in early life are deleterious ( – ) in late life (19). It has also been observed that virtually all humans who live to advanced age have various neoplasms active in their somatic tissue at their time of death. Thus, cancer may not be recorded as their cause of death, but it may have been a significant contributor (20).
Is there genomic chaos during senescence?
Medawar’s prediction that deleterious actions late in life have no evolutionary consequence suggests that in many ways what happens in the genome after reproduction has ceased could essentially be a “free for all”. This notion should be considered in the context of the composition of the genome of complex organisms. The composition of the human genome is only about 1.5% protein coding genes. Compare that to DNA transposons (2.9%), LTR retrotransposons (viral origin, 8.3%), short interspersed elements (13.1%), and long interspersed elements (20.4%) (21). All of these are transposable or mobile elements. When these elements transpose into a coding region during cellular replication, they result in major reductions of fitness. Thus before reproduction ceases, strong purifying selection acts against their activation. However, it is known that transposable genetic elements activation is associated with numerous cancers (22). This is because they are an abundant source of regulatory sequences and therefore capable of activating oncogenes. They are also associated with other age-specific diseases, including autoimmunity, endometriosis, and neurodegeneration (23).
Does Aging Cease?
The genomic chaos I have proposed above would suggest that once individuals have passed their reproductive period, a progression of pathologies resulting from physiological dysregulation associated with the expression of late life deleterious alleles will result in their inexorable death. Or we should expect that the age-specific rate of mortality should always be increasing until all individuals are dead. Yet, the analysis of large cohorts of aging humans (as well as experiments in model organisms, e.g. flies) has shown that late life mortality plateaus exist (24). Several theories were proposed to explain this phenomenon, including density artifacts and population heterogeneity, but these were insufficient. Instead, the existence of this plateau results as a consequence of the force of natural selection falling to very low levels (asymptotically, as opposed to going directly to zero). This was demonstrated experimentally using large cohorts of the fruit fly (25). The issue of course is what circumstances allow an individual to survive the initial acceleration of mortality rates after net future expected reproduction is zero, to make into the mortality plateau phase (or “late-life” as described by Michael Rose and his coworkers.)
Treatments for Aging?
The evolutionary theory of aging provides an ultimate theory that allows us to make sense of its proximate mechanisms. Thus, the evolutionary theory predicts that there will be multiple molecular, cellular, and physiological mechanisms contributing to aging (and eventual death). Attempting to fix any one of them might prolong an individual’s health or life span. Some treatments will only be cosmetic, such as the use of retinol creams in fair-skinned individuals to retard damage to skin caused by UV radiation exposure. Better might simply be to reduce or avoid exposure to sunlight (e.g. don’t sit on beaches to tan.)
In an article I published in 1993, I proposed that the key to retarding the age-specific expression of deleterious alleles must reside somewhere in cellular signaling (26). Specifically given that the only way genetic elements could “know” that they had entered the period past reproduction would rely on their capacity to sense the reproduction of their host. This means that they are inactivated when reproductive hormones are high, and become active when these hormones have dropped to sufficiently low levels. I argued that this explained why hormone replacement therapies (HRT) alleviated some of the symptoms of menopause in human females, as well as reduced some of the physiological measures of aging in those individuals. It is notable that in recent studies of gene expression associated with aging (across taxa) that reduced growth factor signaling is seen, along with down regulation of genes encoding mitochondrial proteins, down regulation of protein synthesis machinery, dysregulation of immune system genes, constitutive responses to stress and DNA damage, dysregulation of gene expression and mRNA processing (27).
It is also notable that one of the best studies of gene expression associated with aging (comparing well characterized populations, resulting from a common ancestor, and with one population undergoing aging, while the other was not), could only find clear differentiation of genes involved in the ABC transporter proteins (28). This system is involved in the transport of a variety of many substrates, and could be involved in the movement of hormones. However, at present there is really no way of knowing. It is also important to recognize that there are ways in which Drosophila is not a good model for gene expression in humans, especially the fact that these insects do not use methylation to the same degree as in gene expression as mammals.
So is treatment for ageing possible? Yes and no. We have had a comprehensive theory of aging (evolution) since the 1940’s. The theory was improved upon greatly at the end of the 20th century. It predicted that there must be multiple levels of aging pathology. The development of advanced sequencing technologies (genomics, transcriptomics, proteomics, metabolomics) at the beginning of the 21st century (the “omics” revolution) has provided tools to understand biological systems at the molecular, cellular, and physiological level (29). CRISPR-Cas-9 methods could be deployed to rewrite patterns of gene expression in specific tissues. Nanomachines now make possible the potential for intervention in cellular processes. These tools can and will be deployed to provide treatments for various aspects of aging, some superficial and others more profound. Unfortunately, such treatments will be beyond the means of most of the world’s people. In the initial phases, these treatments will be reserved for the super wealthy. Of course, since these people are willing to spend billions for joyrides into outer space, they certainly will spend money for the potential to increase both their health and life span. Successful interventions in aging, would drive cost reductions in technology, resulting in wider marketing, but again to those with sufficient means.
Of course, for the majority of the people of the world, the issue isn’t the expansion of their maximum life span, but the realization of their already existing life span. Indeed, health and life span disparity by socially defined race is still a major issue in the Western world. In my book with Alan Goodman we asked the question: “If races are not biological, why are there such persistent differences in health among races?” The response began: “The simple answer to this question is that the social and physical aspects of the environment that contribute to health disparities have not changed enough to eliminate the disparities” (30). So, we are left here with an ethical and moral question, how much of our biomedical research enterprise should be devoted to the extension of the maximum life span of individuals, when so many individuals die early due to exposure to malnutrition, parasitic disease, environmental toxins, and social injustice? Granted, these are not mutually exclusive avenues of research, but quite frankly I would feel better about providing life extension to the rich only when all people are provided with the basic requirements for experiencing healthy and fulfilling lives (e.g., nutrition, shelter, medical care, meaningful employment, social justice).
Joseph L. Graves Jr.
- Graves JL. “A Voice in the Wilderness: A Pioneering Biologist Explains How Evolution Can Help Us Solve Our Biggest Problems”, 2022.
- Kuhn T. “The Structure of Scientific Revolutions” 4th 2012.
- Rockstein M. The biology of aging in insects. In “Topics in the Biology of Aging”, 1966.
- Fischer R.A. “The Genetical Theory of Natural Selection”, 1930.
- R. Arnold and Rose M.R. (Avise J Ed.) “Conceptual Breakthroughs in the Evolutionary Biology of Aging”, 2022.
- Hamilton W.D. 1966. “The moulding of senescence by natural selection”, Theoretical Biology 1966.
- Medawar P. “An Unsolved Problem in Biology”,
- Wattiaux J.M. “Cumulative parental age effects in Drosophila subobscura”, Evolution, 1968.
- Wattiaux J.M. “Parental age effects in Drosophila pseudoobscura”, Experimental Gerontology
- Rose M.R. “Laboratory evolution of postponed senescence in Drosophila melanogaster”, Evolution
- Luckinbill et al. “Selection for delayed senescence in Drosophila melanogaster”, Evolution
- Graves J.L. et al. ”Genomics of Parallel Experimental Evolution in Drosophila”, Molecular Biology & Evolution, 2017.
- Ossowski S. et al. «The rate and molecular spectrum of spontaneous mutations in Arabidopsis thaliana”, Science
- Dönertaş H.M. et al. “Common genetic associations between age-related diseases”, Nature Aging
- Burke M.K. et al. ”Genome-wide analysis of a long-term evolution experiment with Drosophila”, Nature
- Smith K.R. et al. ”Effects of BRCA1 and BRCA2 mutations on female fertility” Proceeding Biological Sciences
- Byars S.G. and Voskarides K. “Antagonistic Pleiotropy in Human Disease”, Journal Molecular Evolution
- Pavard S.E. et al. “Senescence of reproduction may explain adaptive menopause in humans: a test of the “mother” hypothesis”, American Journal Physiological Anthropology
- Stearns S. and Medzhitov R. “Evolutionary Medicine”, 2016.
- Clark, W. “A Means to an End: The Biological Basis of Aging and Death”, 1999.
- Fontdevila, A. “The Dynamic Genome: A Darwinian Approach”, 2011.
- Jang H.S. et al. ”Transposable elements drive widespread expression of oncogenes in human cancers”, Nature Genetics
- Burns K.H. “Our Conflict with Transposable Elements and Its Implications for Human Disease”, Annual Review Pathology
- Rose M.R. “What is Aging?” Frontiers in Genetics
- Rose M.R. et al. “Evolution of late-life mortality in Drosophila melanogaster”, Evolution
- Graves L. “The costs of reproduction and dietary restriction in mammals”, Growth, Development, and Aging 1993.
- Frenk S. and Houseley J. “Gene expression hallmarks of cellular ageing”, 2018.
- Barter T.T. et al. “Drosophila transcriptomics with and without ageing”, 2019.
- Harries L. and Goljanek-Whysall K. “The biology of ageing and the omics revolution”, Biogerontology
- 30. Graves J.L. and Goodman A. “Racism Not, Race: Answers to Frequently Asked Questions, Columbia University Press”, 2022.