Highlights
•
Senescent death of older inviduals could benefit younger kin.
•Theory rules out such programmed death in outbred, dispersed populations.
•C. elegans breed as non-dispersed, clonal populations of protandrous hermaphrodites.
•Under these conditions programmed, adaptive death could evolve.
A widely appreciated conclusion from evolutionary theory is that senescence (aging) is of no adaptive value to the individual that it afflicts. Yet studies of Caenorhabditis elegans and Saccharomyces cerevisiae are increasingly revealing the presence of processes which actively cause senescence and death, leading some biogerontologists to wonder about the established theory. Here we argue that programmed death that increases fitness could occur in C. elegans and S. cerevisiae, and that this is consistent with the classic evolutionary theory of aging. This is because of the special conditions under which these organisms have evolved, particularly the existence of clonal populations with limited dispersal and, in the case of C. elegans, the brevity of the reproductive period caused by protandrous hermaphroditism. Under these conditions, death-promoting mechanisms could promote worm fitness by enhancing inclusive fitness, or worm colony fitness through group selection. Such altruistic, adaptive death is not expected to evolve in organisms with outbred, dispersed populations (e.g. most vertebrate species). The plausibility of adaptive death in C. elegans is supported by computer modelling studies, and new knowledge about the ecology of this species. To support these arguments we also review the biology of adaptive death, and distinguish three forms: consumer sacrifice, biomass sacrifice and defensive sacrifice.
1. Strained relations: C. elegans and the evolutionary theory of aging
Biogerontology, the study of the biology of aging, is a young field. Symptomatic of this is the continuing turnover of central concepts about aging. Here one example is the theory that reactive oxygen species (ROS) are a major cause of aging, and the idea that enhanced antioxidant defense should retard aging, now largely abandoned (Perez et al., 2009; Stuart et al., 2014; Van Raamsdonk and Hekimi, 2010). Another is the idea that retardation of aging by caloric restriction (CR) is a general feature of animals; but effects of CR on mouse lifespan have been shown to be strain dependent (Ingram and de Cabo, 2017; Liao et al., 2010; Mulvey et al., 2014) and results of CR studies on rhesus monkeys (Colman et al., 2009; Mattison et al., 2012) suggest that CR is unlikely to cause major retardation of aging humans in the way that it can do in rodents.
As these central pillars of biogerontology erode and crumble, those remaining become more important to support the field. Critical among these is the evolutionary theory of aging, which provides a powerful explanatory framework for understanding the origins of aging, the natural diversity in aging rate, and the value of aging as an adaptation - or rather, the lack of it (Rose, 1991). An important conclusion from evolutionary theory is that, as a rule, senescence serves no purpose, i.e. it does not contribute to biological fitness (the capacity of an organism to spread its genes in a population). This means that there should not be genes whose naturally selected function is to cause aging.
In this context, the discovery of single gene mutations that dramatically extend lifespan in C. elegans, e.g. those affecting the insulin/IGF-1 signaling (IIS) pathway (Kenyon, 2010), was unexpected. It also caused some discontent with evolutionary biology among some model organism biogerontogists. This has led to a few direct challenges to the classic evolutionary theory (Goldsmith, 2008; Longo et al., 2005; Mitteldorf, 2006; Skulachev, 2002), but more often it manifests in others ways, e.g. as hints that aging is programmed (Budovskaya et al., 2008), and slightly disdainful remarks about the theory, articulated in reviews (Kenyon, 2011) or to interviewers; for example, "[Evolutionary biologists] are always telling you what you can’t think" (Gary Ruvkun) (Anton, 2013).
What is going on here? Is it merely that some model organism geneticists lack understanding of the evolutionary theory? Is this just naive group selectionism on their part? Or is this a case of: there is no smoke without fire? I.e. are there substantive problems in terms of reconciling findings from model organism biogerontology with evolutionary theory? In this essay, we explore the nature of this inter-disciplinary tension, and present a diagnosis of the problem and a possible solution to it. We argue that there is indeed fire, but that the apparent discrepancy with classic evolutionary theory may be resolved by realizing that particular features of C. elegans (e.g. clonality, limited dispersal, protandrous hermaphroditism) are permissive for the natural selection of adaptive death (Dytham and Travis, 2006), i.e. programmed organismal death. Here death can be advantageous due to benefits in terms of inclusive fitness or colony-level fitness. In organisms where adaptive death occurs, promoting senescence and death can be a selected function of genes. In principle, the daf-2 insulin/IGF-1 receptor gene in C. elegans could be such a gene, deserving its nickname of "the grim reaper" (Anton, 2013).
2. The beautiful evolutionary theory of aging
Biogerontologists who dislike the evolutionary theory of aging are missing out, since it is a thing of great explanatory power and beauty. It argues that aging is not itself adaptive, but rather a non-selected evolutionary by-product caused by a decline in natural selection with increasing age (Charlesworth, 1993, 2000, 2001; Hamilton, 1966). Two theories have been proposed for how aging evolves: mutation accumulation (MA) and antagonistic pleiotropy (AP). According to the MA theory, inheritable, late-acting deleterious mutations accumulate in populations, leading to late-life disease and an age increase in death rate (Medawar, 1952). For example, a mutant allele that kills young children will be strongly selected against (as it will not be passed to the next generation), while a lethal mutation with effects confined to people over the age of 80 will experience little selection (people with this mutation will have already passed it to their offspring by that age). The AP theory takes into account the fact that gene mutations can be pleiotropic, i.e. have multiple effects at different places and different times. Where a new allele enhances early-life fitness but promotes disease in later life, it may enhance fitness overall due to the greater impact on fitness of the early effect, in which case frequency of the new allele is likely to increase due to natural selection (Williams, 1957). Thus, although the AP gene variant increases fitness, its promotion of senescence does not (Bourke, 2007).
In terms of the biological mechanisms through which it acts, AP may be viewed in several different ways, as follows. First, as a property of individual structural genes, as in G.C. Williams’ hypothetical example of a gene promoting calcium deposition that in early life enhances bone growth and thereby fitness (e.g. through capacity to run away from predators), but in later life promotes calcium deposition in arteries, leading to cardiovascular disease (Williams, 1957). Next, AP may be viewed as a property of regulatory genes controlling entire programs of growth and differentiation, promoting fitness in early life, but running on in later life in a futile fashion to become pathogenic quasi-programs or pseudo-programs (Blagosklonny, 2013, 2006; de Magalhães, 2012); wild-type daf-2 in C. elegans is a potential example this form of AP (Blagosklonny, 2010; Ezcurra et al., 2018). Third, AP may be viewed as a consequence of biological constraints resulting e.g. from developmental or architectural features of organisms (or Baupläne) (Gould and Lewontin, 1979).
As an illustration of the latter, consider the evolution by natural selection of hermaphroditism in C. elegans from the ancestral gonochoristic state (i.e. females and males) (Ellis and Lin, 2014; Kiontke et al., 2011). The germline of C. elegans hermaphrodites is protandrous, i.e. forming sperm first (∼300) and then oocytes. Sperm are stored in sacs, the spermathecae, through which oocytes pass and are fertilized. Hermaphroditism is thought to have been selected for because it promotes fitness by accelerating population growth rate, allowing C. elegans to outcompete rivals for colonization of transient food sources, typically rotting plant stems and fruit (Schulenburg and Félix, 2017). But as a Bauplan protandry is subject to a major constraint: the switch from sperm to oocyte production initiates egg laying, but it also limits sperm number; once the sperm are depleted, C. eleganscannot reinitiate sperm production. Interestingly, the weakly masculinizing mutation tra-3(e2333) delays the switch from sperm to egg production, thereby markedly increasing brood size which, one might reason, ought to increase fitness. In fact, tra-3(e2333) reduces population growth rate, because the additional time required to make the extra sperm causes a delay in the onset of egg laying (Hodgkin and Barnes, 1991). Thus, tra-3(e2333) exhibits Bauplan-type AP, reducing early life fitness (by delaying the onset of egg laying) but increasing later fitness (by increasing brood size).
Overall, the evolutionary theory of aging provides an explanation for why aging exists and how it evolves. It predicts that aging serves no biological function but is a mere evolutionary epiphenomenon. Aging is pointless and harmful. Consequently, the notion that aging might be a positively selected death program has in the past been vehemently rejected by evolutionary biologists (Austad, 2004; Kirkwood, 2005).
3. Features of C. elegans and S. cerevisiae aging suggest programmed aging
The rigor and explanatory power of the evolutionary theory notwithstanding, some features of aging in certain model organisms seem to suggest some form of adaptive benefit from aging. These include (i) the occurrence of programmed cell death in Saccharomyces cerevisiae, a unicellular organism (Gourlay et al., 2006); (ii) the fact that large increases in lifespan can occur without reductions in fertility (Kenyon, 2005); (iii) the existence of genes which when mutated cause large increases in C. eleganslifespan, particularly genes in the IIS pathway such daf-2 and age-1(phosphatidylinositol-3 kinase), mutation of which can increase worm lifespan as much as 10-fold (Ayyadevara et al., 2008); and (iv) the occurrence of destructive processes promoting senescence within days or even hours of reproductive maturity and with no apparent linked fitness benefit (Labbadia and Morimoto, 2014).
3.1. Defining programmed aging
To explore these issues further it is helpful to clarify the terminology employed. The term programmed aging is somewhat confusing, such that it has even been argued that its use should be discontinued (Bourke, 2007; Rose, 1991). However, we believe that with a little disambiguation and adjustment, the term can be made unambiguous and useful. Its first problem relates to the word aging, which has multiple meanings that are sometimes conflated (Janac et al., 2017). These include benign developmental changes occurring during adulthood (maturation), and age changes that are wholly deteriorative (senescence). In recent years, this usage of senescence is sometimes confused with cellular senescence, a term introduced by Hayflick to describe a particular type of change that affects dividing mammalian cells in vitro (Hayflick and Moorhead, 1961). Cellular senescence is now understood to play a role in development and wound healing as well as organismal senescence (Campisi, 2013; Munoz-Espin et al., 2013). In this essay, when we say aging, we mean senescence in the original meaning of the word.
The concept of programmed senescence views aging as "caused by an ordered series of molecular events which are genetically coded in ways that may be similar to the ways developmental processes are coded" (Johnson and McCaffrey, 1985). Arguably, a shortcoming of this definition is that it is ambiguous as to whether or not programmed here implies a process that contributes to fitness, particularly since the word "programmed" implies purpose (Austad, 2004). To clarify this, the term quasi-programmed was introduced, to denote development-like processes that promote fitness in early life that run on in later life and promote senescence but not fitness (Blagosklonny, 2006); a good example of this in C. elegans is uterine tumour formation (Wang et al., 2018). To reduce ambiguity in this review, we refer to adaptive and non-adaptive programmed aging as programmed and quasi-programmed, respectively. By this terminology, and based on evolutionary theory, the default expectation is that aging in C. elegans may be quasi-programmed, but we are asking: could it actually be programmed? Various observations suggest the latter, and the broader occurrence of programmed aging in certain organisms, as follows.
3.2. Why do unicells exhibit programmed cell death (PCD)?
PCD mechanisms such as apoptosis contribute to fitness in metazoan organisms by various means but not, it is assumed, by promoting organismal death. This implies that in unicellular organisms, where PCD and programmed organismal death (POD) are one and the same, PCD should not exist. Yet in fact budding yeast (S. cerevisiae) do exhibit PCD (Fabrizio et al., 2004; Fabrizio and Longo, 2008; Gourlay et al., 2006). Indeed, the apparent occurrence of POD in budding yeast has been cited in challenges to the view that aging is non-adaptive (Longo et al., 2005; Skulachev, 2002).
3.3. Life extension without reduced fertility
One reason for irritation with the evolutionary theory among model organism biogerontologists is a sense that it led to incorrect conclusions that obstructed the initial development of lifespan genetics (Kenyon, 2011). There is surely some truth to this belief, at least insofar as the classic theory suggests that "genes for aging" should not exist.
The origins of C. elegans lifespan genetics lies with the work of Michael Klass who, in a brilliantly original study, first demonstrated that it is possible to isolate long-lived C. elegans mutants (Klass, 1983). However, Klass interpreted the longevity of the mutants as the result (in most cases) of feeding defects leading to dietary restriction, and concluded that "it is most probable that none of the mutants with increased life spans bear mutations in specific aging genes" (Klass, 1983). With hindsight, it seems strange that the crucial implications of these findings should have been ruled out on this way. Was this the malign influence of the evolutionary theory at work? According to Klass he was well aware of the evolutionary theory when he conducted the study. But his negative conclusions (which were present in the submitted manuscript) were based on his earlier work demonstrating that dietary restriction can increase lifespan in C. elegans (Klass, 1977), and the presence of feeding defects in most of his long-lived mutants (M.R. Klass, personal communication).
Characterising one of Klass's mutants his colleague Tom Johnson subsequently showed that life-extension was not due to dietary restriction. Instead it was shown in a seminal study that a defined mutation, age-1(hx546), dramatically increases both healthspan and lifespan (Friedman and Johnson, 1988). However, again the authors were reluctant to conclude that wild-type age-1 acts in some direct way to promote aging. Instead attention was drawn to an apparent reduction in age-1 mutant fertility, and it was concluded that "It is likely that the action of age-1 in lengthening life results not from eliminating a programmed aging function but rather from reduced hermaphrodite self-fertility or from some other unknown metabolic or physiologic alteration." It was subsequently shown that reducing IIS can increase lifespan without reducing fertility (Gems et al., 1998; Johnson et al., 1993). Moreover, preventing hermaphrodite reproduction altogether does not increase lifespan (Friedman and Johnson, 1988; Kenyon et al., 1993; Klass, 1977). Subsequently, dissociation of effects on lifespan and fertility have been demonstrated in a variety of contexts (Grandison et al., 2009; Kenyon, 2004, 2005; Piper et al., 2017).
3.4. Why does IIS shorten lifespan?
Mutation of daf-2 can greatly increase C. elegans lifespan (Kenyon et al., 1993). This fact taken alone suggests that daf-2 has evolved to cause death, which somehow promotes fitness. Because evolutionary theory argues against this, various attempts have been made to explain away this apparent contradiction to it. One proposal was that daf-2 controls the switch between development to adulthood and dauer larva developmental arrest. Since dauer larvae are much longer lived that wild-type adults (Klass and Hirsh, 1976), it was suggested that the longevity of daf-2 mutant adults is attributable to mis-expression of dauer longevity-assurance mechanisms in the adult (Kenyon et al., 1993), and comparisons of gene expression profiles of dauers and daf-2 mutants supported this idea (McElwee et al., 2003, 2004). However, more recently it was discovered that daf-2 mutant longevity can be dissociated from such dauer-ness (Ewald et al., 2015). Another suggested solution is that IIS is a conserved, nutrient-sensitive regulator of trade offs between reproduction and somatic maintenance (survival), enabling life history plasticity in the face of a changing nutrient environment (Partridge and Gems, 2002). Yet it appears increasingly doubtful that inadequate somatic maintenance (e.g. prevention and repair of stochastic molecular damage) is a major cause of aging in C. elegans (Ezcurra et al., 2018; Gems and Partridge, 2013). A third possibility, explored here, is that by promoting death daf-2 somehow promotes fitness.
3.5. Why does C. elegans possess active mechanisms of self-destruction?
The last decade has seen discoveries that further strain the plausibility of the claim that C. elegans aging is non-adaptive. Most strikingly, it was discovered that mechanisms that protect protein folding homeostasis undergo a major collapse shortly after reproductive maturity. This collapse is detectable by day 3 of adulthood as a dramatic decline in heat stress resistance, and an ∼80% reduction in induction after stress challenge of genes involved in the heat shock and unfolded protein responses (Ben-Zvi et al., 2009; Shemesh et al., 2013; Taylor and Dillin, 2013), followed by increased levels of protein aggregation (Ben-Zvi et al., 2009; David et al., 2010). More recently, it was shown that this collapse occurs, startlingly, within 6 h of reproductive maturity, and is promoted by signaling from the germline (Labbadia and Morimoto, 2015, 2014). The reports on the proteostatic collapse phenomenon present no evidence of a coupled increase in fitness; but when interviewed a key senior author of these studies (R.I. Morimoto) remarked: "I absolutely believe that aging is programmed" (Anton, 2013).
Does this mean quasi-programmed or actually programmed? In principle, it could be either. One possibility is that proteostatic collapse is somehow coupled to individual fitness, e.g. by reducing the constraints imposed by protein quality control on protein synthesis rate (Sherman and Qian, 2013) or secretion (Labbadia and Morimoto, 2014) (i.e. quantity over quality), thereby increasing yolk production capacity - and fitness. Here the collapse process is quasi-programmed AP, i.e. proteostatic collapse does not in itself promote fitness but is merely an unselected by-product. To be precise, according to this scenario reducing protein folding capacity has two consequences: increasing yolk production (programmed) and proteostatic collapse (quasi-programmed). Another possibility is that promotion of death promotes fitness though inclusive fitness effects (detailed below); here the process is programmed and not AP, i.e. death itself actually promotes fitness, rather than being an unselected by-product of some other trait that does. Several authors of these studies view both possibilities as plausible (J. Labbadia and R.I. Morimoto, personal communication).
A further explanation that has been suggested to explain proteostatic collapse draws on the disposable soma theory. This theory argues that aging is caused by accumulation of stochastic molecular damage, and that fitness is maximised by partition of available resources (e.g. caloric) optimally between reproduction and somatic maintenance, in a manner sufficient only to assure longevity during the reproductive period (Kirkwood, 1977). Thus, down-regulating maintenance of proteostasis could increase resources available for reproduction at the cost of reduced protection of the soma, and reduced lifespan (Ben-Zvi et al., 2009; Labbadia and Morimoto, 2015, 2014; Shai et al., 2014). (Arguably, increasing fidelity of protein synthesis by slowing translation rate, as noted above, is not a somatic maintenance mechanism in this sense). However, it remains unclear that stochastic molecular damage is a major cause of senescence in C. elegans, and other proximate mechanisms have been identified (Ezcurra et al., 2018; Gems and Partridge, 2013; Van Raamsdonk and Hekimi, 2010).
Since the discovery of proteostatic collapse in 2009, more studies have appeared documenting other examples of early deterioration in C. elegans. For example, long-term memory is impaired by days 2–3 of adulthood and is entirely lost by day 5 (Kauffman et al., 2010), and degeneration of the dendrites of PVD neurons occurs from ∼day 4–7 that is actively promoted by the antimicrobial peptide NLP-29 and driven by autophagy (E et al., 2018). Also, in early adulthood C. elegans neurons extrude membrane-bound vesicles full of aggregated proteins and organelles (exophers), suggesting pathological upheaval at this time (Melentijevic et al., 2017). More broadly, analysis of the temporal dynamics of development of a range of senescent pathologies showed concerted disease development largely occurring from day 4–12 of adulthood (Ezcurra et al., 2018). These pathologies include intestinal atrophy and yolk steatosis, which are coupled in an IIS-promoted process of gut-to-yolk biomass conversion, suggesting quasi-programmed AP action in this case (Ezcurra et al., 2018). In addition, some features of organismal death suggest programmed or quasi-programmed function, such as waves of cellular necrosis promoted by calcium influx accompanying rigor mortis and intestinal death (Coburn et al., 2013; Galimov et al., 2018).
Altogether, one can see how these results might seem to constitute some kind of challenge to the classic evolutionary theory of aging, and certainly explain the ambivalent attitude to it among model organism biogerontologists. Although all these phenomena may eventually be explicable in terms of AP, they also suggest that it is worth considering more carefully the possibility that death might somehow promote fitness in C. elegans and S. cerevisiae.
.....
R E S T A T S O U R C E : ScienceDirect
Edited by Engadin, 14 May 2019 - 09:20 PM.
