The hyperfunction theory: an emerging paradigm for the biology of aging

The process of senescence (aging) is predominantly determined by the action of wild-type genes. For most organisms, this does not reflect any adaptive function that senescence serves, but rather evolutionary effects of declining selection against genes with deleterious effects later in life. To understand aging requires an account of how evolutionary mechanisms give rise to pathogenic gene action and late-life disease, that integrates evolutionary (ultimate) and mechanistic (proximate) causes into a single explanation. A well-supported evolutionary explanation by G.C. Williams argues that senescence can evolve due to pleiotropic effects of alleles with antagonistic effects on fitness and late-life health (antagonistic pleiotropy, AP). What has remained unclear is how gene action gives rise to late-life disease pathophysiology. One ultimate-proximate account is T.B.L. Kirkwood’s disposable soma theory. Based on the hypothesis that stochastic molecular damage causes senescence, this reasons that aging is coupled to reproductive fitness due to preferential investment of resources into reproduction, rather than somatic maintenance. An alternative and more recent ultimate-proximate theory argues that aging is largely caused by programmatic, developmental-type mechanisms. Here ideas about AP and programmatic aging are reviewed, particularly those of M.V. Blagosklonny (the hyperfunction theory) and J.P. de Magalhães (the developmental theory), and their capacity to make sense of diverse experimental findings is assessed.

the proximate mechanisms of antagonistic pleiotropy, an important evolutionary-genetic principle in the biology of aging (Williams, 1957).
Devising effective explanations of senescence as a whole is a challenge that has, unfortunately, defeated biogerontologists. Why is aging so difficult to understand? Arguably, there are two reasons in particular. First, because it is a multifactorial process, as reflected in the fact that diseases of aging are largely multifactorial in terms of their etiology. Thus, any individual theory about a given senescent etiology will be insufficient to explain the whole phenomenon. Second, because the relative contribution of the factors that together cause aging is variable and context dependent, differing according to the senescent pathology considered, to environmental conditions and type of organism (Ezcurra et al., 2018;Gems and de Magalhães, 2021).
Though this presents a difficult challenge, it does not mean that senescence is impossible to understand. Though aging is a multifactorial process, a plausible hope is that the number of individual factors involved is relatively small, particularly when viewed in terms of broad determinative principles. Moreover, given the common evolutionary origins of animal species, these principles are likely to be operative in diverse organisms. Thus, the challenge for biogerontology is to understand these individual principles, and how they combine to cause aging-related disease, and aging as a whole (Gems and de Magalhães, 2021).
While attempts to understand aging from the 1990s onwards were strongly focused on the role of accumulation of molecular damage as a cause of aging, earlier accounts tended to be more broadly multifactorial (Comfort, 1979;Cutler, 1984;Dilman, 1994;Finch, 1990;Strehler, 1977). Multifactorial models, particularly that of Vladimir Dilman, will be discussed later in this article. In the meantime, it should be emphasized that programmatic mechanisms and antagonistic pleiotropy (AP) as described here are elements of a wider, multifactorial process.
In the following discussion I will first introduce the concept of AP, as described by George Williams, and then go on to describe attempts to explain AP in terms of actual biological mechanisms (e.g. biochemical, cellular, developmental) that form a pathophysiological basis of the process of senescence. These include Tom Kirkwood's disposable soma theory, Misha Blagosklonny's hyperfunction theory, João Pedro de Magalhães's developmental theory, and Vladimir Dilman's ontogenetic theory. The latter three theories share key features, and can be referred to collectively as the programmatic theory. For definitions of key terms, see the glossary.

The evolution of aging
Diseases of aging are strange in that their principal etiology is not any of the usual suspects that disrupt normal function (infectious pathogens, mechanical injury, mutation, toxins etc), but rather the process of evolution. Natural selection favors individuals that leave the largest number of surviving offspring, not necessarily those that live the longest. With advancing age after the onset of reproduction, natural selection weakens with the result that new alleles that impair health in later life are more likely to escape selection and accumulate DNA replication errors to glycation of proteins (Holliday, 1997;Rattan, 2008), but particular attention was paid to damage arising from reactive oxygen species (ROS) such as the superoxide (O 2 -) free radical produced as a by-product of oxidative metabolism (Beckman and Ames, 1998). If molecular damage causes aging it follows that the level of cellular maintenance that prevents damage accumulation is an important determinant of aging rate, and that the genetic determination of lifespan is likely to involve regulation of cellular maintenance mechanisms (Holliday, 1997;Partridge and Gems, 2006). In short, aging rate reflects the balance between stochastic damage accumulation and somatic maintenance levels (the damage/maintenance paradigm) (Fig. 1B).
The observation that cellular maintenance processes are costly in resource terms led to the deduction that trade-offs might exist between cellular maintenance and reproductive effort. Given limited resource availability in the wild, optimization of the trade-off would inevitably lead to levels of cellular maintenance below those necessary to entirely prevent damage accumulation. The result of this is a rapidly aging, disposable soma, in contrast to the immortal, non-aging germline (Kirkwood, 2005;Kirkwood, 1977) (Fig. 1C).
The disposable soma (DS) theory provides a clear and logical account of how trade-offs could arise between fitness traits at different points in the life history: early benefit (increased reproduction) and later cost (increased mortality due to gradual damage accumulation). Though not originally conceived as such (Armstrong, 2019) it provides a cogent explanation for the proximate mechanisms of AP gene action (Kirkwood and Rose, 1991). Yet despite its elegance, the DS theory remains unproven, and some experimental findings argue against it (Grandison et al., 2009;Piper et al., 2017;Zajitschek et al., 2019). One limitation is its dependence upon the damage/maintenance paradigm. While molecular damage, DNA damage in particular (Schumacher et al., 2021), is certainly a primary causal mechanism in some forms of senescent pathology (e.g. cancer), for many others this is far from clear (see below). While it is possible that DS effects could exacerbate pathologies where damage is a major driver, their importance as a major cause of aging currently seems unlikely (discussed further below) (Blagosklonny, 2007b;Blagosklonny, 2010e). But while its explanatory utility remains uncertain, as the first ultimate-proximate model, the DS theory was important in providing a template for the development of ultimate-proximate theories.

Ultimate-proximate theory 2. Where aging is programmatic
In the early 2000s it became increasingly clear that the insulin/IGF-1 signaling (IIS)/ mammalian (or mechanistic) target of rapamycin (mTOR) network can control aging rate across a range of organisms Kenyon, 2010). Discovering the IIS/mTOR-regulated, downstream mechanisms of aging itself promised to provide fundamental insights into the nature of senescence. Guided by contemporary beliefs, early attempts to understand this guessed that wild-type IIS accelerated aging by reducing somatic maintenance (e.g. antioxidant defense) thereby increasing reproductive output, consistent with the DS theory (Fontana et al., 2010;Gems and McElwee, 2005;Holzenberger et al., 2003;Honda and Honda, 1999;Kirkwood, 2005;Lithgow and Walker, 2002;Partridge and Gems, 2002b;Vanfleteren, 1993) (Fig. 1D). However, experimental findings during the same period were starting to raise doubts about the assumption that molecular damage is a primary cause of aging, particularly oxidative damage (Bjelakovic et al., 2007;Chong et al., 2007;Howes, 2006;Keaney and Gems, 2003;Keaney et al., 2004).
Against this background, two biogerontologists independently saw a way to rearrange existing concepts and findings into a new ultimate-proximate theory that is very different to DS. Although initial accounts of the theory, put forward by João Pedro de Magalhães (de Magalhães and Church, 2005) and Mikhail (Misha) Blagosklonny (Blagosklonny, 2006a), differ in emphasis and in detail, they have the same conceptual core (the programmatic theory). de Magalhães and Blagosklonny each put forward many ideas, some more persuasive than others; the following account of the framework of ideas centered on the programmatic theory is to a certain extent my own synthesis of their ideas.

The core programmatic theory
Growth hormone (GH), IIS and mTOR are part of a nutrient-sensitive signaling network that promotes growth, development and aging. This suggests that growth and development somehow cause aging. But three lines of reasoning initially seemed to argue against this deduction. First, it suggests that there is a program for aging, implying that aging is a purposeful adaptation; but according to evolutionary theory aging is non-adaptive, and therefore not programmed (Austad, 2004;Kirkwood, 2005;Partridge and Gems, 2002b). Second, growth and development are not obviously linked to somatic maintenance and molecular damage accumulation, which was assumed to be the main cause of senescence.
Third, it is difficult to see how growth processes can limit lifespan. The programmatic theory provides answers to each of these three objections.
First, the program problem. Here it is helpful to consider Williams' own initial thoughts about how AP might work at the level of gene function, which involve a hypothetical gene for calcium deposition. A new allele appears that increases Ca 2+ deposition during bone development, and thereby promotes fitness (e.g. by aiding escape from predators); in later life, its continued action promotes Ca 2+ deposition into blood vessel walls, contributing to arteriosclerosis ( Fig. 2A) (Williams, 1957). In principle, a mechanism could evolve to switch off calcium deposition, but due to the selection shadow it does not. Here, an evolved function of a gene continues or runs on in a futile fashion in later life causing pathology. The principle involved here is notably different to the DS theory. If this can occur for the function of structural genes of this sort, why not for regulatory genes controlling entire developmental programs and involving action of large numbers of regulated genes ( Fig.  2B)?
Evolutionary theory rules out biological programs for aging (i.e. aging as an adaptation) for most species, but not programmatic or program-like mechanisms that promote senescence (Blagosklonny, 2007c;de Magalhães and Church, 2005). Part of the confusion arises because the word program here has two meanings. On the one hand it can mean a genetically-determined and complex process involving changes in gene expression, cellular function, tissue composition etc (programmed in the mechanistic sense); on the other hand it can mean a program for something that promotes fitness (programmed in the adaptive sense) in later life programmatic changes occur that are programmed in the mechanistic but not the adaptive sense, or as Blagosklonny describes them, quasi-programmed (Blagosklonny, 2006a). The derived noun quasi-program, describing a programmatic, pathogenic entity, is particularly helpful for discussion of programmatic pathophysiology.
To explain the quasi-program concept, Blagosklonny employs various homely analogies. For example, one wants hot water for tea and so boils some in a saucepan. If, having taken the water, one leaves the saucepan on the hot stove, the pan will become damaged.
Here a purposeful program for preparing hot water becomes a futile quasi-program for damaging the saucepan (Blagosklonny, 2006a). In living organisms later-life off switches for developmental programs are often absent because of declining selection in later life (Blagosklonny, 2006a;Williams, 1957). Similarly, de Magalhães uses the analogy of a program to build a house, and imagines a carpet layer working on pointlessly after completion of the program: "ever-increasing layers of carpets will eventually prevent doors from opening, and ultimately, nobody will be able to get in or out of the house" (de Magalhães, 2012). This starts to explain how growth promotion by IIS/mTOR can promote aging: they promote both developmental programs and senescence-promoting quasi-programs (Blagosklonny, 2006a;de Magalhães and Church, 2005).
But what about molecular damage? de Magalhães and Blagosklonny both suggest that the assumption that aging is largely and primarily caused by accumulation of molecular damage is incorrect (Blagosklonny, 2006a;de Magalhães and Church, 2005). As Blagosklonny articulates it, the primary mechanisms of aging do not involve loss of function, but rather the opposite: too much function or, as he puts it, hyperfunction, driven by wild-type gene action. This claim challenges the entrenched assumption that aging is fundamentally a passive process of system failure and breakdown. Thinking of mTOR effects in particular, Blagosklonny argues that it is the opposite: an active, self-destructive process.
Finally, how can processes of growth and development lead to disease? Broadly speaking, the programmatic theory argues that developmental changes in adulthood, including latelife continuation of developmental programs, is pathogenic, causing disruption of tissue and organ function. Several examples of simple forms of developmental run-on follow.
Presbyopia is a type of long-sightedness that increases with age. Its cause is continued, futile growth of the lens during adulthood, leading to a gradual increase in lens thickness (Strenk et al., 2005). In men, the prostate gland typically exhibits a gradual increase in size as a response to long term exposure to dihydrotestosterone. This leads eventually to benign prostatic hyperplasia (Nacusi and Tindall, 2011;Waters et al., 2000) and increased risk of prostate cancer, a major cause of age-related death in men (Untergasser et al., 2005). In a third example, this time hypothetical, the process of synaptic pruning in the brain that promotes cognitive development runs on in later life, leading to age-related cognitive decline (De Magalhaes and Sandberg, 2005). A final example affects babirusas, a type of wild pig found in Indonesia. Adult males have curved, tusk-like maxillary canine teeth that point backwards up over the snout. These continue to grow until in some cases the backward curve of their growth trajectory drives them into the cranium, sometimes piercing it through (Macdonald, 2018). Here, in each case, continuation of normal, wild-type growth leads to pathology. However, most aging-related diseases are etiologically multi-factorial, and contributions of programmatic etiologies more complex (described below).
A simplified scheme of the programmatic theory is shown in Fig. 2C. Beyond these core tenets, Blagosklonny and de Magalhães have each contributed additional ideas, which extend and strengthen the theorem. For example, Blagosklonny elaborates upon how quasiprograms affect lifespan and disease, and makes pointed critiques of various concepts arising from the damage/maintenance paradigm, while de Magalhães integrates the programmatic theory conceptually with earlier developmental theories of aging, and explores evolutionary mechanisms beyond AP and IIS/mTOR that contribute to programmatic aging. Ideas from these two commentators will be discussed in turn.

Blagosklonny's research approach
To understand Blagosklonny's thinking it is helpful to know something about his background and approach. Unlike most biogerontologists, he has a clinical background (M.D. in internal medicine, Ph.D. in experimental medicine and cardiology). He is a proponent and practitioner of conceptual research, which uses explorations of the scientific literature to develop new hypotheses and to test them, sometimes repurposing existing data to draw new conclusions in ways unintended by the original researchers (Blagosklonny, 2007d;Blagosklonny and Pardee, 2002). He initially applied this approach for research in experimental oncology, including a focus on anti-cancer drug action (Blagosklonny, 2005a;Blagosklonny, 2005b;Blagosklonny, 2005c). An interest in the potential of the drug rapamycin as an anti-cancer agent (Blagosklonny and Darzynkiewicz, 2002) led him to the idea that this drug might act by inhibiting aging. This led in turn to the question of the causes of aging itself, and a re-examination of central ideas in biogerontology, both ultimate (e.g. antagonistic pleiotropy, disposable soma) and proximate (e.g. molecular damage accumulation, telomere shortening).
A strength of conceptual research is the relative ease of combining findings and concepts from different scientific disciplines, for example between basic and clinical research fields (Blagosklonny, 2003b), enabling broader insights. Blagosklonny's conclusions draw extensively from both biogerontological sources and, importantly, clinical literature on age-related disease; thus, his ideas are more strongly grounded in contemporary research findings (including clinical findings) than in traditional theories about aging. They are also influenced by earlier ideas developed in Russia in the 1970-80s (Dilman, 1994) (discussed further below).
Given that the starting point of the work was to understand the action of rapamycin, his perspective is highly focused on this drug and the protein that it inhibits, mTOR (Blagosklonny, 2007a;Blagosklonny, 2010b;Blagosklonny, 2012c;Blagosklonny, 2014b;Blagosklonny, 2019a), which restricts its scope somewhat. However, his arguments are often applicable (and applied) to the broader endocrine and signalling network that regulates development and growth, including GH and IIS. In many of his discussions, the aging process that he describes is the set of processes controlled by GH/IIS/mTOR which, as he sometimes emphasises, is not the entire aging process.
Key to the initial development of Blagosklonny's theorem was his realization that growth stimulation leads rapidly to cellular senescence when the cell cycle is arrested, with no involvement of molecular damage (Blagosklonny, 2003a). (Note here that cellular senescence, as defined by Hayflick, should not be confused with the broader phenomenon of senescence). His main ideas are set out in three early essays (Blagosklonny, 2006a;Blagosklonny, 2007b;Blagosklonny, 2008b), particularly the first of these. He then produced a very long series of follow-up essays (over 60), some derived from additional conceptual research with newer findings, and some responding to individual new papers. Many of these essays argue that an mTOR hyperfunction model better explains published findings than the conventional damage-based view, and often include explanatory reiterations of his version of the programmatic theory. The repetition to an extent reflects the lack of response to his work from other biogerontologists; "Repetitio est mater studiorum" as he notes at one point (Blagosklonny, 2018b). Some present new elaborations and additions to the theory, e.g. (Blagosklonny, 2007c;Blagosklonny, 2009b;Blagosklonny, 2010c;Blagosklonny, 2010d;Blagosklonny, 2018a;Blagosklonny, 2019b). For brief summaries of the content of 68 of his essays, see Table 1. The essays draw to some extent on his own laboratory work, mostly on the regulation of cellular senescence, see e.g. (Demidenko et al., 2009a;Demidenko and Blagosklonny, 2008;Demidenko and Blagosklonny, 2009;Demidenko et al., 2010;Demidenko et al., 2009b;Leontieva and Blagosklonny, 2016;Leontieva et al., 2015;Leontieva et al., 2012a;Leontieva et al., 2012b). At times they seem written in haste, as if painted with rough brushstrokes, but what they sometimes lack in polish they often make up for in originality and creative insight. Key elements of Blagosklonny's framework of ideas, beyond the core programmatic theory include the following.

How do quasi-programs promoted by IIS/mTOR limit lifespan?
Aging is a deteriorative process that causes decline and death, but how? The core programmatic AP account provides a cause, but does not explain how it leads to life-limiting harm. It is here that Blagosklonny in particular sheds light, helped by three basic ideas that combine to give a very different view of aging to that which is traditional in biogerontology. First, that death due to aging is caused by disease. Second, that aging results from a nested series of life-limiting etiologies. Third, that programmatic AP promoted by IIS/mTOR causes life-limiting disease in a largely non-cell autonomous manner.

5.2.1
Aging is pathology-Biogerontologists usually use lifespan or mortality rate as metrics of aging rather than senescent pathology, and tend to view aging as a process distinct from age-related disease (Gems, 2015;Hayflick, 2007). By contrast, influenced by his medical training, Blagosklonny sees aging very much in terms of diseases: "No one dies from healthy senescence: humans and other mammals die from senescenceassociated diseases such as cancer, stroke, myocardial infarction, [etc]" (Blagosklonny, 2006a). He argues that biogerontologists often have an erroneous view of lifespan as a function of damage, and diseases of aging as relatively incidental ( Fig. 3A) (Blagosklonny, 2007b;Blagosklonny, 2012a). Surveying clinical literature that is typically beyond the biogerontological purview, he notes that many senescent pathologies and conditions are linked to mTOR over-activity, including cancer, atherosclerosis, hypertension, type 2 diabetes, osteoporosis, osteoarthritis, macular degeneration, and Alzheimer's and Parkinson's diseases (Blagosklonny, 2006a;Tsang et al., 2007). Thus, programmatic AP limits life by causing diseases of aging.

Aging as a nested series of life-limiting etiologies-
A tenet of the DS theory is that aging as a whole is caused by molecular damage accumulation, suggesting that enhanced somatic maintenance could prevent aging altogether (de Grey et al., 2002;Kirkwood and Rose, 1991). There do exist organisms for which this appears to be true: in the filamentous fungus Podospora anserina, reducing mitochondrial ROS production can prevent senescence entirely and produce a seemingly non-aging state (Dufour et al., 2000). By contrast, Blagosklonny views mTOR's impact on lifespan not as an effect on aging as a whole but rather of promotion of multiple life-limiting pathologies (or multimorbidity). He reasons that if mTOR hyperfunction were suppressed (e.g. by inhibition with rapamycin) then other disease etiologies would become more life limiting (e.g. other quasi-programs, oxidative damage, mutation, telomere shortening) (Blagosklonny, 2006a;Blagosklonny, 2007b;Blagosklonny, 2021b) Inhibition of IIS/mTOR can increase lifespan substantially, suggesting an effect on aging as whole and, therefore, the existence of an overall aging process. An alternative view is that quasi-programs can arise independently in multiple tissues, promoting diverse senescent pathologies, all of which are supported by IIS/mTOR since they are fundamental drivers of growth and development (Fig. 3C). Thus, IIS/mTOR promotes senescent multimorbidity, but not the entire aging process. This suggests that the idea of "aging as a whole" was, in some sense, a conceptual Fata Morgana in the minds of biogerontologists produced by a combination of IIS/mTOR inhibition effects and theories such as DS.

Non-cell autonomous action of quasi-programs-
The idea that aging is caused by molecular damage suggests that senescence originates largely as a cell autonomous process, in which impairment of basic cell function leads to cell breakdown, and from there upward to tissue and organ failure. Hence, aging is a problem of cellular metabolism and cellular maintenance. By contrast, in the mTOR quasi-program model basic cell function is in many cases unimpaired; instead, cell function is changed in a way that disrupts tissue and organismal function. Pathology results from loss of tissue, organ and physiological homeostasis rather than of cellular homeostasis. Similarly, de Magalhães observes that one can broadly divide gene function into that specifying basic cell biology (respiration, DNA replication, repair) and that specifying developmental programs, and argues that the main causes of aging are located in the latter and not, as the damage/ maintenance paradigm assumes, in the former (de Magalhães and Church, 2005).
Pathogenic effects of cellular hyperfunction are largely cell non-autonomous, involving hyperplasia, dysplasia, hypertrophy, hyper-secretion, atrophy and cell signal resistance (see below). Blagosklonny describes how many diseases of aging are caused by hyperfunction (Blagosklonny, 2006a;Blagosklonny, 2012a;Blagosklonny, 2013d). As one example, myocardial infarction results from a combination of factors, including coronary atherosclerosis and arterial spasm, hypertension, myocardial hypertrophy and thrombosis, for each of which hyperfunction is a contributory factor. Proliferation and hypertrophy of arterial smooth muscle cells (SMC) and lipid accumulation contributes to atherosclerosis; proliferation, hypertrophy, elevated contractile function and hyperstimulation of arterial SMC all contribute to hypertension; mTOR hyperfunction contributes to cardiac hypertrophy; and increased aggregation and adhesion of platelets contributes to increased thrombosis. An example of hyperfunction causing atrophy is progression of osteoporosis due to osteoclast hyperactivity, which is promoted by mTOR (Blagosklonny, 2006a). Thus, the pathogenic effects of hyperfunction in the main do not work from bottom up, by harming individual cells that exhibit it; instead, hyperfunctional cells are robust and, in themselves, healthy, even though they might be damaged (as in cancer cells) (Blagosklonny, 2008b;Blagosklonny, 2012a). Instead, hyperfunctional cells cause pathology at the tissue and organ level. However, one noxious cell-autonomous effect of mTOR hyperfunction is reduced autophagic clearance of aggregation-prone proteins, increasing risk of several age-related neurodegenerative diseases (Blagosklonny, 2006a).
If pathogenic effects of cellular hyperfunction are largely cell non-autonomous, then what about aging in single cell organisms? Notably, in budding yeast (Saccharomyces cerevisiae) inhibition of TOR can increase lifespan (Blagosklonny and Hall, 2009), implying that TOR promotes cell death in this species. Blagosklonny suggests that aging in bacteria results from damage accumulation (Blagosklonny, 2012a), and presumably this applies to unicellular eukaryotes too.

Programmatic etiologies as a target for intervention-This new model
provides an account of rapamycin action against aging. Quasi-programs occur due to the late-life selection shadow, which has resulted in insufficient off switches (Williams, 1957); rapamycin in later life can rectify some of this insufficiency (Blagosklonny, 2006a). More broadly, the model predicts a general preventative approach to age-related disease: to allow enactment of programs but, in later life, inhibit quasi-programs. This possibility is implicit in Williams' calcium gene scenario: one could imagine a treatment to prevent it from promoting vascular calcification in later life.
Validating anti-aging drugs requires biomarkers of aging, the search for which has been a long-standing objective in aging research. This includes a major program by the US National Institute on Aging initiated in 1988 to try to identify biomarkers of aging in rodents (Warner, 2004). Since aging is a developmentally-driven multimorbidity (rather than a metabolic Gems Page 10 Ageing Res Rev. Author manuscript; available in PMC 2022 January 11. problem within cells), then a good biomarker is progression of multiple diseases of aging (Blagosklonny, 2009b). Potential anti-aging drugs can be identified through re-analysis of outcomes of drug tests against individual diseases, to look for wider anti-disease effects. For example, rapamycin administered to renal transplant patients unexpectedly reduced cancer rates in several studies (Blagosklonny, 2009b). This principle implies that obesity accelerates aging, since it increases the rates of multiple aging-related diseases (Blagosklonny, 2011b), meaning that reducing obesity is an anti-aging treatment.

5.2.5
Pathogenic effects of mTOR acting as a brake on growth-A recurrent theme in Blagosklonny's thinking is that mTOR is a driver of growth that becomes pathogenic when growth has to cease (Fig. 4A,B). mTOR itself can act as a brake on growth by cutting off upstream growth-promoting signals via feedback inhibition. Specifically, mTOR acts through S6 kinase to inhibit IRS-1 (insulin receptor substrate) (Fig. 4C), leading to receptor signal resistance, including insulin resistance that contributes to type 2 diabetes (Blagosklonny, 2006a). Signal resistance caused by mTOR may also contribute to age-related loss of stem cell function (Blagosklonny, 2008a).
Blagosklonny views mTOR within the cell as a microcosm at the cellular level of homeostatic function at the organismal level, particularly in the hypothalamus, where pathology also develops due to signal resistance (Blagosklonny, 2013b;Dilman, 1971). As he points out, like the hypothalamus, mTOR integrates diverse signals, in the latter from insulin, mitogens, cytokines, oxygen, and nutrients. The recurrence of AP behavior of mTOR suggests the presence of a trade-off between the risks of allowing vs preventing growth that hinges on mTOR activity. The idea that a brake on growth at the end of ontogenesis causes aging was proposed previously, in a different context (Bidder, 1932).

5.2.6
Programmatic loosening as a feature of aging-According to programmatic theory, the processes of development and aging unfold as the result of shared developmental mechanisms. But compared to development proper, aging is far more variable and, seemingly, disordered. Development is a highly precise, stereotyped process. By contrast, each person's individual experience of aging is unique; one starts with lower backpain and shingles, and progresses to macular degeneration and death from colon cancer, while another starts with osteoarthritis in the fingers and urge incontinence, and moves on to vascular Gems Page 11 Ageing Res Rev. Author manuscript; available in PMC 2022 January 11. dementia and an eventual lethal stroke. To some extent this variation will reflect differences in genotype and environment. The importance of factors other than genetic difference is illustrated by the considerable degree of discordance even between monozygotic twins in terms of diseases of aging, ~50% for type 2 diabetes, for example (Poulsen et al., 1999). Even in C. elegans where population cohorts of genetically identical individuals can be maintained in the same Petri dish, the lifespans of individual nematodes vary greatly (from ~10-30 days under standard culture conditions), and different causes of death can be observed (Rea et al., 2005;Zhao et al., 2017).
Variability in outcome would seem to be an inherent property of the aging process. Blagosklonny argues that variability of outcome is a characteristic feature of quasiprograms: "As an unintended continuation of the developmental program, the quasi-program is not very precise" (Blagosklonny, 2007b). Consistent with this, chronological age is a good surrogate for biological age (as estimated using a methylation clock, DNAm age, described below) during development but increasingly less so in later life (Horvath and Raj, 2018). Moreover, gene expression heterogeneity remains low during development but increases during aging (Isıldak et al., 2020). The authors of that study interpreted this as evidence that stochastic rather than developmental processes contribute to aging; an alternative interpretation is that such heterogeneity reflects developmental stochasticity characteristic of quasi-programs.
The progressive developmental loosening associated with quasi-programs may result from the age decline in natural selection. In earlier life greater selection maintains the precision of developmental programs, but it wanes with increasing age, this precision declines. A formal possibility is that genetic determinants of precision are subject to antagonistic pleiotropy involving trade-offs between earlier developmental precision and later precision loss.
In summary, the programmatic theory defines an etiological principle that contributes to many diseases of aging. Wild-type gene action promotes pathogenic quasi-programs in later life, in line with Williams' original conception of AP. Promotion of quasi-programs by wild-type levels of IIS/mTOR can at least partially explain how they accelerate aging. This conceptual framework also provides a principle for treatment (inhibition of quasiprograms), actual possible treatments (e.g. mTOR inhibitors), and approaches to test them experimentally in humans.

Blagosklonny on the attack
In many of his essays, Blagosklonny uses conceptual research to compare how well damagebased and programmatic theories do at explaining findings past and present. Overall, these find in favor of the latter, and support critiques of several traditional biogerontological claims and concepts, which he sometimes pokes fun at for their perceived inadequacies. 2003; Keaney et al., 2004). The crisis in the ROS theory that culminated in 2008-09 has been reviewed extensively elsewhere (Gems and Doonan, 2009;Perez et al., 2009;Shields et al., 2021;Van Raamsdonk and Hekimi, 2010). It is true that there is a correlation between aging and accumulation of molecular damage. Of this, Blagosklonny says: "With age, I have accumulated a considerable knowledge regarding the ROS theory. Does this knowledge cause my aging? I hope not" (Blagosklonny, 2008b). Instead, he argues that molecular damage accumulation is symptomatic of later stages of diseases of aging caused by hyperfunction.

Against disposable soma-
The mTOR-driven quasi-programs (mTOR/QP) model provides an alternative ultimate-proximate model to the disposable soma (DS) theory, that is closer to Williams' own original idea about proximate mechanisms of AP.
Blagosklonny even refers to this model as "disposable soma theory 2", insofar as the soma is disposable as the result of mTOR/QP action (rather than resource allocation etc) due to AP and the later-life selection shadow (Blagosklonny, 2013c). He points out that the mTOR/QP model originates from consideration of diverse scientific findings, unlike either Harman's free radical theory (Blagosklonny, 2008b) or the DS theory (Blagosklonny, 2010e), and argues that it outperforms DS at explaining a range of phenomena.
For example, the life-extending effects of dietary restriction (DR) have been reconciled with DS by the allocation hypothesis. Here it is argued that DR leads to diversion of resources from reproduction to somatic maintenance, thereby increasing lifespan (Holliday, 1989;Kirkwood and Shanley, 2005;Masoro and Austad, 1996) (Fig. 5A). Blagosklonny argues that this is implausible for a number of reasons. Perhaps most cogently, a more parsimonious prediction from DS is that increasing resources for somatic maintenance would increase lifespan and vice versa (Blagosklonny, 2007b). By contrast the mTOR/QP model argues simply that reduction in nutrients reduces mTOR activity, which suppresses both growth and QPs (Fig. 5B).
As another example, inhibition of protein biosynthetic genes in C. elegans increases lifespan, and it was suggested that this is due to freeing up of resources that are then invested in somatic maintenance (Hansen et al., 2007;Hipkiss, 2007). Blagosklonny argues instead that reducing protein synthesis prevents QP progression, pointing out that mTOR promotes protein synthesis (Blagosklonny, 2007b). As a third example, it was argued from the DS theory that women live longer due to increased investment into somatic maintenance (Kirkwood, 2010). Blagosklonny reasons that DS predicts that women should live shorter, since they invest more into reproduction; he suggests instead that increased mTOR activity promotes muscle growth in males, but accelerates aging (Blagosklonny, 2010a;Blagosklonny, 2010d;Leontieva et al., 2012b).

5.3.3
Against stress-response hormesis-For some toxins, a lower dose range exists that promotes health (e.g. increased growth rate, stress resistance), an effect known as hormesis (Calabrese et al., 1999). In some cases this occurs due to induced expression of cytoprotective genes, so called stress-response hormesis (Gems and Partridge, 2008). For example, sub-lethal heat stress induces a heat shock response that can protect C. elegans against subsequent heat stress that would otherwise be lethal (Lithgow et al., 1995).
Treatments that chronically increase ROS levels can increase C. elegans lifespan (Schulz et al., 2007;Yang and Hekimi, 2010), seemingly a challenge to the ROS theory. However, it has been suggested that this could reflect hormesis, where induction of antioxidant defences by ROS protects against aging caused by ROS; this includes ROS production by mitochondria inducing mitohormesis (Radak et al., 2005;Ristow and Zarse, 2010). Blagosklonny sees this account as an attempt to rescue the ROS theory from findings that contradict it, and sends up its logical awkwardness (increasing ROS reduces ROS). He compares it to Baron Munchausen's adventure when, falling into a swamp, he escapes by pulling himself up by his own hair (Blagosklonny, 2008b). He suggests instead that effects of stress on lifespan more plausibly result from inhibition of growth processes that drive both development and quasi-programs (Blagosklonny, 2007b).
A constraint to progress in biogerontology is the ever-increasing volume of research output, which means it is difficult for researchers to keep abreast of the whole field. For this reason, the conceptual research approach is timely. By using it Blagosklonny has been able to develop a coherent and empirically well-supported framework of understanding of many aspects of the biology of aging. Yet it by no means presents a full picture of the process of aging, nor does it pretend to.

João Pedro de Magalhães
The programmatic theory reconciles a mechanistic theory based on developmental change with the AP theory, and provides an account of IIS/mTOR action on aging. But both developmental changes during aging, and AP are wider phenomena, as has been emphasized (Maklakov and Chapman, 2019), and all causes of aging is a wider category still. de Magalhães, originally from Portugal and now at the University of Liverpool, gives some consideration to programmatic etiologies in the broader sense (de Magalhães and Church, 2005), and explores issues relating to the programmatic theory that Blagosklonny touches on more briefly or not at all, including the following. How does the programmatic theory relate to earlier developmental theories of aging? And: how does it relate to epigenetic changes during adulthood?

From developmental theories to the programmatic theory
Over the years it has often been suggested that aging is a form of developmental abnormality, resulting e.g. from increased or decreased gene silencing, transcriptional deregulation, dysdifferentiation or overdifferentiation (de Magalhães and Church, 2005). Certain broad characteristics of aging are suggestive of its developmental nature. For example, different mammals (e.g. mice vs humans) differ greatly in terms of rate of development, timing of maturity, and of appearance of many of the same diseases of aging (e.g. cancer, osteoporosis, cataracts) (de Magalhães and Church, 2005). This is also true of the gradual age-dependent decline in cognitive function (De Magalhaes and Sandberg, 2005). Moreover, across mammalian species the ratio between time from conception to sexual maturity and adult lifespan is approximately constant, at about 1:4 (Charnov, 1993), suggesting a possible mechanistic relationship between development rate and adult lifespan (de Magalhães and Church, 2005). Similarly, the complex and stereotyped nature of many manifestations of senescence (e.g. atherosclerosis, osteoporosis, male pattern baldness) seem more consistent with programmatic pathophysiology than random molecular damage (Blagosklonny, 2006a).
While it is clear that both stochastic damage and programmatic changes contribute to senescence, in terms of their relative importance these are competing theories. The former predominated during recent decades, but less so prior to that. For example, as de Magalhães points out (de Magalhães, 2012), the discovery of effects of DR on aging, initially in fruit flies and rats, arose from the idea that slow growth rate results in slower aging (McCay and Crowell, 1934;Northrop, 1917;Osborne et al., 1917;Park, 2010), including Bidder's hypothesis (Bidder, 1932). Later work showing that although life-long DR limits growth, adult-limited DR was sufficient to extend lifespan in fully grown mice (Weindruch and Walford, 1982) appeared to argue against the developmental theory; however, it is consistent with stunting of quasi-programs during adulthood.
Similarly, in chico (insulin receptor substrate) mutant fruit flies with reduced IIS, increased lifespan is seen in both chico homozygotes and heterozygotes, but only the homozygotes exhibit reduced body size (Clancy et al., 2001;Tu et al., 2002). Thus, chico mutant longevity is not caused by their reduced body size , suggesting that reduced growth does not slow aging; however, these results, again, do not rule out suppression of quasi-programs during adulthood. In fact, chico -/+ heterozygote females show reduced egg production (Clancy et al., 2001), consistent with biosynthetic (and programmatic) insufficiency.
Arguably, the confusion here stems partly from ambiguities arising from the words growth and development. Simply understood these refer to changes occurring during ontogenesis that largely cease at sexual maturity. A metric of growth is adult body size, and to speak of growth after maturation in organisms with determinate growth, after they have stopped growing, makes little sense. But in the programmatic model, growth pathways promote quasi-program progression, or "twisted growth" as it has been described (Blagosklonny, 2008b) during adulthood. Similarly, development. A quasi-program is defined as "a program for development that has not been turned off" (Blagosklonny, 2006a). Yet it is clear that what is meant here is development in the sense of all complex programs involving growth and differentiative change: not only those of ontogenesis, but also for example those involved in reproduction and tissue repair and remodelling during adulthood (c.f. the role of "senescent" fibroblasts in wound healing) (Demaria et al., 2014). To avoid this confusion the term programmatic theory is used here, rather than developmental theory as previously (Maklakov and Chapman, 2019). This avoids confusion with earlier developmental theories of aging, and over the meaning of development, and also alludes to the reconciliation of ultimate and proximate perspectives in the theory.

From the programmatic theory to epigenetic aging and biological clocks
A recent topic of interest among biogerontologists has been epigenetic changes during adulthood (Benayoun et al., 2015;Horvath and Raj, 2018). de Magalhães notes studies showing similarities between changes in gene expression occurring during development and aging (Lui et al., 2010;Somel et al., 2010;Takasugi, 2011) as evidence that drivers of epigenetic change are programmatic rather than stochastic (de Magalhães, 2012). As a developmental process, quasi-program execution is expected to involve epigenetic change. Thus, the programmatic theory provides an explanation in terms of ultimate and proximate causes for epigenetic changes in aging (though this does not rule out the action of other contributory mechanisms).
The question of how biological time is marked during aging is central to biogerontology, and many forms of clock have been proposed. These include a metabolic clock in the rate-ofliving theory (Pearl, 1928), a telomere shortening replicometer in replicative senescence (Olovnikov, 1996), and epigenetic changes such as DNA methylation (Horvath, 2013;Horvath and Raj, 2018). According to the programmatic theory, the major aging clock is a developmental one. de Magalhães sees epigenetic changes as likely to arise from the ticking of a developmental clock (de Magalhães, 2012), whose speed is increased by IIS/mTOR (de Magalhães and Church, 2005). Consistent with this, the methylation clock is slowed by conditions in which mTOR is inhibited (Blagosklonny, 2018a), for example in mice treated with rapamycin, in long-lived Ames (Prop1), Snell (Pit1), and Laron (Ghr) dwarf mutant mice with defective GH signaling (Cole et al., 2017;Consortium, 2021;Petkovich et al., 2017;Wang et al., 2017), and in rapamycin-treated cultured human fibroblasts and keratinocytes Matsuyama et al., 2019). Moreover, mTOR functions in circadian clocks, underscoring the role of mTOR in setting the pace of biological time (Blagosklonny, 2018a).
The developmental basis of epigenetic changes during adulthood in mammals has received further support from recent work by Steve Horvath and colleagues. Of particular note, methylation clocks mark time during embryogenesis and postnatal development as well as (running much more slowly) during adulthood, and clock methylation sites are strongly associated with genes specifying developmental processes. The latter include those under regulation of the polycomb repressor complex (PRC), including multiple Hox genes, leading to the conclusion that "the essence of the aging process itself is an integral part of, and the consequence of the development of life" (Raj and Horvath, 2020). The link between clock methylation sites and PRC-regulated genes was also recently seen in a universal mammalian methylation clock, implying evolutionary conservation of developmental clock mechanisms (Consortium, 2021).
Almost a century ago, Raymond Pearl attributed rate-of-living effects, as seen in effects of ambient temperature on lifespan in poikilotherms (e.g. Drosophila, C. elegans) (Klass, 1977;Loeb and Northrop, 1917) to metabolic rate (Pearl, 1928), an idea that was subsequently linked to the ROS theory (Sohal and Weindruch, 1996). The programmatic theory suggests instead that effects of temperature on both development and aging reflects a change in the rate of developmental processes.
by Vladimir Dilman, but modified here in the light of recent advances. Dilman, who died in 1994, was a preeminent gerontologist and clinician in the U.S.S.R., in many ways the Soviet Union's equivalent to thoughtful biogerontologists in the West such as Alex Comfort, Bernard Strehler, Richard Cutler and Caleb Finch, but his theories are little known outside Russia. He was also the father of Misha Blagosklonny.

Dilman's ontogenetic and four models theories
Dilman argues that, overall, aging and age-related disease are etiologically multifactorial, and attributable to four distinct disease models: ecological, genetic, accumulational and ontogenetic (Dilman, 1994) (Chapter 11) (Fig. 6A). Ecological here is in the sense used by Frederic Ludwig to describe the main disease model in medicine, where diseases result from harmful extrinsic factors ranging from infectious pathogens to dietary toxins (Ludwig, 1980). Genetic here refers to inherited genetic diseases, rather than somatic mutations or pathogenic action of wild-type genes. The accumulational model includes molecular damage accumulation, but also other types of pathogenic accumulative processes. The ontogenetic model involves the continued action of developmental processes in later life promoting senescence.
According to this four models scheme, mean lifespan and inter-individual variation in lifespan is largely attributable to ecological and genetic determinants, and maximum lifespan to accumulational and, particularly, ontogenetic determinants. While Dilman's ontogenetic model clearly prefigures the programmatic theory, it was derived as a generalization from a now partially outdated theory about the control of aging by the hypothalamus. Briefly, this argued that aging is largely caused by programmed loss of hypothalamic sensitivity to feedback inhibition, leading to compensatory endocrine changes with pathogenic consequences, what he called hyperadaptosis (Dilman, 1984).
The ontogenetic processes that he particularly refers to are those controlling systemic homeostasis in adulthood, namely energy metabolism (insulin), reproduction (estrogen) and adaptive systems (glucocorticoids).
Yet in amongst Dilman's many ideas and reflections are key elements of the new programmatic theory, including the following. The ontogenetic model itself is prescient; as Dilman describes it: "the genetic program of an organism is not constructed according to the rule 'from beginning to end,' but proceeds 'from a beginning, with the end unspecified'" (Dilman, 1994) (Chapter 6). He notes in passing the congruence between the ontogenetic model and antagonistic pleiotropy (Dilman, 1986). Like Blagosklonny, he emphatically rejects the distinction drawn between normal aging and age-related disease: "normal aging is a disease, or more precisely, a sum of diseases" (Dilman, 1994) (Chapter 6). Like de Magalhães, he emphasizes that ontogenetic etiologies are operative at a level above that of the cellular; however, he views the key level as systemic, while in the modern programmatic theory it is at all supra-cellular levels (from tissue microenvironment to systemic). By contrast, at the subcellular level, accumulative mechanisms are more important (e.g. DNA damage accumulation).
Again like de Magalhães, he views the fixity among mammals of the length of the developmental period as a proportion of total lifespan as support for the role of ontogenetic Gems Page 17 Ageing Res Rev. Author manuscript; available in PMC 2022 January 11. mechanisms in aging. Here he makes an interesting observation. The standard explanation for this constancy is that an increased selection shadow (e.g. due to increased extrinsic mortality) leads to the evolution of earlier sexual maturity and shorter lifespan, via independent mechanisms (earlier reproduction increases fitness, earlier selection shadow allows earlier senescence to evolve) (Charnov, 1993;Harvey and Purvis, 1999;Promislow, 1993). But Dilman wonders about the obverse situation, where longevity evolves to increase fitness via a longer reproductive span. He asks: given the independent mechanisms model, why should a delay in sexual maturity occur? More plausibly, longevity evolves through deceleration of the entire ontogenetic program, which slows both development and aging (Dilman, 1994) (Chapter 11); Blagosklonny later made a similar point (Blagosklonny, 2013a). According to this interpretation, the longer development time in humans relative to other higher primates evolved not due to any fitness benefit, but as a side-effect of selection for increased longevity, i.e. it is an unselected outcome of antagonistic pleiotropy, an evolutionary spandrel (Gould, 1997;Gould and Lewontin, 1979). Consistent with the ontogenetic model is the earlier puberty, accelerated cognitive decline and shorter lifespan in mice over-expressing growth hormone, and the converse effects in dwarf and DR mice (Chandrashekar et al., 2004).
A major question raised by multifactorial models of aging is how the component factors interact with one another and here Dilman is, again, insightful. He argues that both ecological and accumulative factors can modulate ontogenetic aging, and that accumulational and ontogenetic mechanisms, though fundamentally independent are yet interwoven. He also suggests that stochastic damage may occur either in a wholly probabilistic fashion (as in somatic DNA damage accumulation), or as the result of genetically-determined programmatic mechanisms, what he refers to as "regular stochastic processes". Here he has in mind ROS generated by mitochondrial metabolism. A recent case fitting this description is protein aggregation in C. elegans, resulting from a programmatic down-regulation of proteostatic mechanisms from around the time of sexual maturation (Labbadia and Morimoto, 2014).

A new multifactorial model based on Dilman's four models
While Dilman's ontogenetic theory is a predecessor of the programmatic theory, one can, in a rather recursive fashion, modify and update his four models theory in the light of its newer descendent (Fig. 6B). First, by acknowledging the predominance of the wild-type genotype as the determinant and driver of programmatic mechanisms. This incorporates AP as a major determinative mechanism. Second, to unify ecological and genetic determinants into a single, broad category of causes apart from wild-type function. These represent the major causes of diseases in earlier life, where normal, wild-type function is disrupted. The rearranged scheme emphasizes the centrality of wild-type function, which can be illustrated by a thought experiment. If ecological and genetic factors were, by magic, eliminated, aging would still occur, and there would be little change in either the rate of acceleration of mortality with age (Gompertzian aging) or maximum lifespan. By contrast, if mechanisms of aging from wild-type function were removed, Gompertzian aging would be greatly reduced, though not, one assumes, entirely, due to accumulative mechanisms independent of wild-type function.
The role of inherited mutations in aging in this model requires further explanation. One proposed cause of aging, and mechanism of its evolution, is inherited mutations whose deleterious effects are expressed only later in life, as exemplified by Huntington's disease (Haldane, 1941;Medawar, 1952). An unexplained feature of this mutation accumulation theory is the mechanism that causes a mutation to remain harmless throughout development and early adulthood, and then become harmful in later life. One possibility is that it is wild-type gene action and programmatic changes that cause the deleterious effects of such mutations to become expressed, or unmasked. Returning to the thought experiment again: according to the unmasking hypothesis, if one eliminated the wild-type gene action component from the revised Dilman model (Fig. 6B), a carrier of the Huntington mutation would not develop the disease.
An important feature of multifactorial models of aging, noted previously (Ezcurra et al., 2018;Gems and de Magalhães, 2021), is their highly context-dependent nature. For a given animal species, the relative contribution of different etiological factors will vary not only between different elements of senescence (e.g. different diseases of aging), but also between individuals. Moreover, the relative contribution of different factors may vary more broadly between species. For example, in humans, environmental, genetic, ontogenetic and accumulative factors all contribute to the development of atherosclerosis, a major determinant of age-related death in humans. The relative importance of these factors will vary between individuals. In one, a major cause may be obesity, and increased cholesterol accumulation; in another, genetic predisposition (e.g. the APOE4 mutation), in a third, wild-type gene action may act alone. By contrast, the large uterine tumors that are invariably present in senescent C. elegans develop due to wholly programmatic mechanisms (Wang et al., 2018a;Wang et al., 2018b), consistent with the view that plastic, programmatic mechanisms play an exceptionally large role in aging in C. elegans Lohr et al., 2019).

The ontogenesis of the programmatic theory
Scientific discovery is rightly the object of wonder, something to marvel at. In my view, this applies to the programmatic theory. In the early 18th century a fierce priority dispute broke out between Isaac Newton and Gottfried Leibniz arguing over who invented calculus. The modern view is that this was silly, and that both men developed it independently. Such independent, simultaneous discoveries are quite common, and illustrate how scientific innovation is not so much a reflection of individual genius, as interactions of imaginative individuals with a collective and evolving fabric of knowledge. Once the fashion among scientists, these days I-said-it-first-ism risks being interpreted as a character weakness (principally, egotism).
The transpersonal nature of scientific thinking applies well to the programmatic theory, which has evolved over a long period, with certain key contributors. Dilman's overall scheme, based around his hypothalamic model, was flawed and did not cohere well, yet included a number of prescient elements which contributed to Blagosklonny's more compelling and better-supported theorem. Dilman himself noted similarities between his ontogenetic theory and Bidder's hypothesis, formulated in the 1930s (Dilman, 1986).

Gems Page 19
Ageing Res Rev. Author manuscript; available in PMC 2022 January 11. By contrast, de Magalhães' version of the programmatic theory was developed without knowledge of either Dilman or Blagosklonny's work. The possibility of the programmatic theory was, as it were, hanging in the air from the early 2000s, and both de Magalhães and Blagosklonny spotted it, with Blagosklonny developing it in more detail with respect to its capacity to explain the origins of disease. Yet the programmatic theory of the mid-2000s forms only part of a wider biology of aging, as discussed here and elsewhere Lohr et al., 2019;Maklakov and Chapman, 2019). Of Dilman: he was in many ways ahead of his time and his work deserves greater recognition than it received during his lifetime (at least, outside Russia). He was noted for his creative and integrative thinking style (Napalkov, 2001), characteristics shared by Blagosklonny, and de Magalhães too.

Concluding remarks
The programmatic theory of aging as developed by Blagosklonny and de Magalhães promises to eventually serve as part of a general framework of understanding aging and the pathophysiology of late-life disease. This could at long last provide the field of biogerontology with an effective explanatory paradigm similar to that provided by the germ theory for the study of infectious disease, and the periodic table for chemistry. More work is needed to explore, test and expand this framework of ideas. This includes elements not discussed here, which will be the subjects of future essays, such as the concept of hypofunction (Maklakov and Chapman, 2019), biological constraint, how quasi-programs are initiated, and how quasi-programs interact in disease pathophysiology.

Glossary
Adaptive cellular senescence A type of programmed cellular senescence, e.g. where "senescent" fibroblasts contribute to wound healing (Demaria et al., 2014). This contrasts with quasi-programmed, non-adaptive cellular senescence. In some cases cellular senescence may have both adaptive and futile elements, as when a cell leaves the cell cycle due to DNA damage (programmed, adaptive) and then becomes hypertrophic and hypersecretory (quasi-programmed, non-adaptive).

Antagonistic pleiotropy (AP)
Where action of a given gene is both beneficial and detrimental to fitness. If the latter occurs later in life and is therefore subject to weaker selection, such a gene may be favored by natural selection, and promote aging (Williams, 1957).

Damage/maintenance paradigm
Theory that aging is largely caused by accumulation of molecular damage, which can be prevented by somatic maintenance functions. Various theories of aging are based on this broad assumption.

Disposable soma
Theory proposing that natural selection favors investment of limited resources into reproduction rather than somatic maintenance, accelerating damage accumulation and, therefore, senescence (Kirkwood, 1977).

Hyperfunction
Where wild-type gene action and biological function actively leads to senescent pathology, as opposed to passive random damage or wear and tear (Blagosklonny, 2006a).

Onion model (new term)
This views aging as a nested series of life-limiting pathologies (Blagosklonny, 2006a;de Magalhães, 2012). Thus, lifespan is a function not of the overall aging process, but of those processes generating life-limiting pathologies. Which pathologies are life-limiting depends upon environment and animal species.

Programmed aging
Senescence caused by a relatively ordered series of biological processes that promotes fitness via inclusive fitness or group fitness. Evolutionary theory predicts that this does not occur in most animal species, but can in some cases (Galimov and Gems, 2021;Lohr et al., 2019).

Programmatic aging
Where senescence is genetically determined, including where complex biological processes contribute to senescence, but not necessarily to fitness (cf. quasi-programs, costly programs). Senescence caused by a relatively ordered series of biological processes that does not promote fitness; may occur due to futile run-on of wild-type programs that promote fitness earlier in life (Blagosklonny, 2006a).

Run-on
Futile continuation of gene function or processes in later life, leading to pathology (de la Guardia et al., 2016) (cf. quasi-program).

Selection shadow
Decrease in selection with increasing age, leading to weaker selection against genes with deleterious effects on fitness and health the later in life those effects are expressed. Environmental factors that increase mortality rate (e.g. predators, infectious pathogens, starvation) can deepen the selection shadow.

Senescence
The overall process of deterioration with age or the resulting pathological condition (not to be confused with cellular senescence (sensu Hayflick), which is a particular form of cell cycle arrest affecting some vertebrate cell types). Although aging has several meanings, in the biological context it is usually synonymous with senescence.

Ultimate-proximate theories of aging
These combine explanations of the evolutionary (ultimate) and mechanistic (proximate) causes of aging into a single integrated account.

Gems Page 30
Ageing Res Rev. Author manuscript; available in PMC 2022 January 11.

Fig. 1. Ultimate and proximate mechanisms of aging (traditional interpretations).
A, Antagonistic pleiotropy. A new allele that causes a fitness benefit in early life but a fitness cost (e.g. increased pathology) in later life may cause a net benefit in overall fitness due to the selection shadow (Williams, 1957). B, The damage/maintenance paradigm. Aging is caused by accumulation of stochastic molecular damage, whose levels can be controlled by somatic maintenance functions. C, The disposable soma theory. Investment of resources in reproduction more than somatic maintenance can increase fitness due to the selection shadow (Kirkwood, 1977). Genes promoting such resource allocation will exhibit antagonistic pleiotropy (Partridge and Gems, 2002a). D, Regulation of aging by nutrient pathways. Traditional view based on damage/maintenance paradigm (Partridge and Gems, 2006). The hyperfunction model argues that it is in fact the growth, development function that plays the main role in promoting aging.

Gems Page 31
Ageing Res Rev. Author manuscript; available in PMC 2022 January 11.

Fig. 3. Concepts from the broader programmatic theory.
A, The false distinction between aging and disease; adapted from (Blagosklonny, 2007b). Left, aging as a process distinct from late-life disease; here disease is incidental to understanding aging. Right, aging as diseases/pathologies caused by wild-type gene action; here disease/pathology is critical to understanding aging. Accounts of aging in terms of damage/maintenance tend to neglect the importance of disease in the aging process (Blagosklonny, 2012a;Gems and de Magalhães, 2021). B, The onion model of the relationship between aging, disease and lifespan (Blagosklonny, 2006a;de Magalhães, Gems Page 33 Ageing Res Rev. Author manuscript; available in PMC 2022 January 11. 2012). Lifespan is a function of one or more specific life-limiting pathologies, rather than any underlying process of aging as a whole. Left, lifespan is limited by pathology a. Right, a life-extending treatment acts by eliminating a and b. Lifespan is now limited by pathology c. C, How single genes can exert large effects on aging. Growth control pathways act in diverse tissues and organs, driving diverse programs and quasi-programs, the latter leading to diverse pathologies. 2 diabetes). D, Evolution of adaptive, programmed cellular senescence by exaptation from non-adaptive quasi-programmed cellular senescence (hypothetical scheme). Most cellular senescence is pathogenic, caused ultimately by AP. But some forms of cellular senescence have evolved a function, as in wound healing promoted by senescent fibroblasts (Demaria et al., 2014). In the context of adaptive cellular senescence, the term "senescence" is illogical.  A, The allocation hypothesis, based on the disposable soma theory. Reduced food leads to increased investment in somatic maintenance. B, Explanation based on the hyperfunction theory. Here reduced nutrients reduces growth pathway signaling, thus reducing quasiprogrammed hyperfunction. Figure redrawn from (Blagosklonny, 2007b). The latter is both more parsimonious and consistent with existing knowledge of pathophysiology of various diseases of aging.  A, Dilman's four models mechanism. This representation is based on his account, but leaves out his hypothalamic threshold model (Dilman, 1994). Here the major determinant of aging and mean and, particularly, maximum lifespan is ontogenetic mechanisms, with accumulative mechanisms (e.g. molecular damage accumulation) playing a lesser role. Ecological (i.e. extrinsic) and genetic determinants contribute to senescent pathology, and are a major determinant of mean lifespan, and variation in individual lifespan. B, Dilman's multifactorial model rearranged, to incorporate more recent ideas. Here the major determinant of programmatic (ontogenetic) senescence is wild-type genotype, including pathogenic wild-type gene action due to antagonistic pleiotropy. But genotype-specified programmatic mechanisms contributing to aging clearly involve more than AP (Maklakov and Chapman, 2019). The model makes clear the centrality of wild-type function in aging.
Even in the absence of other factors (ecological, genetic), senescence will occur, and maximum lifespan will change little. Factors that contribute to disease throughout life and do not reflect wild-type function, both extrinsic (ecological) and intrinsic (mutation) are grouped into one broad category. Included here is mechanical damage (mechanical senescence), not included by Dilman. One determinant not shown is the effect of extrinsic factors on damage accumulation (e.g. skin damage due to solar UV radiation).

Gems
Page 38 Ageing Res Rev. Author manuscript; available in PMC 2022 January 11.

Gems
Page 39 Recaps main theory using an analogy. Aging is not like a rusting car, but one without brakes. TOR is the engine, driving pathogenic quasi-programs, rapamycin a brake. Concise overview of the biology of mTOR and its role in aging, including a greater emphasis on effects in budding yeast.
Response to report in Aging that resveratrol inhibits S6 kinase. Resveratrol effects on aging may work through mTOR pathway. Rapamycin action on aging is an unintended effect of a growth inhibiting antifungal; a similar, "side effect" type explanation for resveratrol action on aging is more plausible than the xenohormesis hypothesis.
Recapitulates arguments that DR acts by inhibiting mTOR, and arguments against the molecular damage and allocation hypotheses.
Why human lifespan is rapidly increasing: solving "longevity riddle" with "revealed-slow-aging" hypothesis. Blagosklonny  Argues that mTOR activity is higher in men, and reiterates Dilman's theory that menopause is due to a quasi-program: run-on of the decline in hypothalamic sensitivity to inhibition by estrogen that triggers puberty by allowing FSH production. Later hypothalamic insensitivity leads to FSH over-production, causing a futile acceleration of follicular atresia.
Argues instead that mTOR and IIS pathway genes exhibit AP, and rapamycin is a reversed-AP drug.
Introduces term post-aging syndrome for senescence where mTOR hyperfunction is suppressed.
Why the disposable soma theory cannot explain why women live longer and why we age. Commentary on finding that metformin increases mortality rate in young adult male mice. Elaboration of "revealed slow aging" hypothesis (Aging 2010 2 177).