William R. Clark, Department of Molecular, Cell and Developmental Biology, University of California, Los Angeles, California, 90024

The evolutionary theory of senescence is based largely on principles outlined by Williams in 1957, and consists of two relatively independent parts. The first part builds on ideas first put forward by Medawar, Haldane and others, to explain how something as negative as senescence could have been positively selected in evolution, particularly since most animals in the wild do not reach an age where senescence is expressed. Williams proposed that the genes responsible for the negative effects of senecence (senescence effector genes) were fixed in evolution by a process he called antagonistic pleiotropy, wherein a subset of genes selected because they confer a reproductive advantage early in life may have harmful effects in the post-reproductive period; negative selection against these harmful effects fails because, as pointed out by Medawar, the force of natural selection declines with age. The evolutionary history of senescence-causing genes is seen as a nondirected accumulation of genes selected on a basis independent of senescence per se. In the second portion of his paper, Williams made a series of predictions about how the age of organisms at reproductive maturity, fecundity, lifespan and the timing of the onset of senescence would all interact in the life history of a species. These latter predictions, which do not depend at all on details of the mechanisms of selection of senescence effector genes, have been validated by numerous experiments over the past several decades. On the other hand, it has become increasingly evident that the senescence effector genes did not, as would be predicted by antagonistic pleiotropy, accumulate in a random, non-directed fashion in various species over evolutionary time. Rather, everything we know about these genes suggests they were present in eukaryotic founder cells shortly after, or even congruent with, the emergence of eukaryotes from their prokaryotic ancestors, and have been stringently conserved ever since. Complicated explanations of how so-called "death genes" may have evolved in eukaryotes are thus not required. It is suggested that the evolutionary theory of senescence should be focused on those evolutionary principles that have been validated experimentally, and that the notion of antagonistic pleiotropy be dropped from theories of the evolution of senescence.

Some fifty years ago, Peter Medawar launched a series of lectures and discussions that led to what is generally considered the "modern era" of thinking about the evolution of senescence. Medawar summarized his early ruminations in his classic treatise, An Unsolved Problem of Biology (Medawar, 1952). The problem, as Medawar and others (e.g., Haldane, 1941) saw it, was how to explain the evolution of something as negative to the interests of the individual reproductive organism as senescence, which has death as its implicit endpoint. This is especially difficult to visualize since, in the wild, the majority of members of any given species do not live long enough for senescence ever to be expressed. Aside from humans, where senescence is now a major factor in death, senescence is largely seen only in animals kept in zoos or reared in laboratories, where they can be protected from accidental causes of death such as predation, starvation, or physical trauma. Most animals - in some species, the vast majority - die of accidental causes well before senescence begins to be expressed. How could a trait not expressed by the majority of members of a species be acted on by natural selection, and come to be fixed in the species as a whole?

Medawar was among the first to recognize what has become a foundation stone of contemporary evolutionary theories of aging: the declining force of natural selection with age of the individual. Harmful genetic events that are expressed prior to the reproductive period in an animal’s life history will be strongly selected against, whereas the expression of such genes at later stages will not be subject to negative selection. Medawar speculated that spontaneously arising variants of genes that display a harmful effect only later in life - senescence effector alleles - would simply accumulate in a species over evolutionary time; the allelic variants of the genes responsible for Huntington’s chorea and Alzheimer’s disease are often cited as examples. He also raised the possibility that late-acting, harmful genetic alterations could arise in senescence regulator genes, as well as in the senescence effector genes themselves. This concept of senescence regulator genes was not followed up for several decades after Medawar proposed it, but may be his most important contribution to thinking about the evolution of senescence.

These ideas about the origin of genes causing senescence were carried forward a few years later in a seminal paper by George Williams (Williams, 1957). Williams’ paper actually contains two mechanistically independent parts. In the first part, Williams essentially restates and refines slightly the previous ideas of Haldane, Medawar and others about the origin and evolution of genes responsible for senescence. In this context, Williams contributed the term antagonistic pleiotropy to the lexicon of discussions about the evolution of aging. This notion suggests that allelic variants of genes involved in promoting senescence would likely have been positively selected on the basis of whether or not they enhance an individual’s ability to survive until the reproductive period, and/or to carry out reproductive activities in a successful fashion. Some of these variants may, by random chance, have negative effects on the organism at later stages in life, but since there would presumably be no reproductive consequences of these negative effects, natural selection cannot operate to suppress transmission of these alleles to subsequent generations. Given that every allele in any genome can only be positively selected on the basis of its contribution to reproductive fitness, this is a relatively modest extension of previous thinking on the subject. Nevertheless, it is obvious that Williams, like Medawar, felt that an accretion of such late-acting mutations could account for the overall "program" of senescence seen in various species today.

Quite apart from his speculations on the evolutionary origin of genes responsible for senescence (Medawar’s distinction between senescence effector genes and senescence regulator genes is lost in Williams’ 1957 paper), Williams also made a series of important predictions about how reproductive timing and accidental death rates should influence fecundity, senescence and lifespan. By and large these predictions, which have stimulated a great deal of fruitful research in senescence in the past four decades, are not dependent on his ideas about the origins of senescence effector genes. Nevertheless, both sets of ideas are generally included in what is now called the "evolutionary theory of senescence" (hereinafter referred to simply as the evolutionary theory). Most of the early attempts to test the validity of the evolutionary theory were directed to those aspects dealing with the interactions between senescence and species mortality rates, or between senescence and reproductive maturity and fecundity. Such studies, which relied mostly on classical transmission and population genetics, (admirably summarized in Rose, 1991), have largely validated Williams’ predictions.

Most of the elements of the evolutionary theory were laid down at a time when molecular genetics was in its infancy, and it fails to account for a number of features of senescence that have become increasingly apparent in recent years. In particular, there are difficulties with that portion of the theory implying that senescence effector genes gradually accrued in different species over evolutionary time, in a random process not directed toward aging per se. The problem is that the senescence-related genes described so far do not behave that way at all. Recent genetic analyses have shown that the fundamental mechanisms of senescence, and the genes underlying them, are remarkably similar in virtually every eukaryotic organism studied, which is hardly consistent with a random, independent accumulation of late-acting harmful mutations over the evolutionary history of eukaryotes. Senescence does not behave like a gradually accruing genetic pattern, either in terms of fixation of the underlying genes, or of the lack of any apparent random genetic drift that might be expected in a genetically passive and unregulated phenomenon.

Thus, while the initial proposals of Haldane, Medawar, Williams and others offered a way out of a seemingly irresolvable dilemma about the origin and evolution of senescence-related genes, this portion of the resulting evolutionary theory suffers from serious genetic, molecular, and even evolutionary drawbacks. Curtsinger et al. (1994), in studying how stable polymorphisms might arise in large populations, concluded that antagonistic pleiotropy probably plays a limited role in explaining persistence of genetic variation in fitness components. They did not, however, propose explanations for the selection of senescence-related genes. I propose that instead of gradually accruing over long periods of time in a random fashion, nearly all of the genetic elements of senescence - the genes that cause senescence (senescence effector genes), and those that oppose its effects (senescence resistor genes) - were set in place very shortly after, or in some cases even before, the emergence of eukaryotes from their prokaryotic ancestors. These elements have been rigorously conserved through subsequent evolution, with very little change, and thus discussions of how they may have been selected in spite of harmful effects directed toward the individual, and how they could have been selected in individuals not living long enough to express them, must be shifted back to the context of the very earliest stages of eukaryotic phylogeny. A strong case can be made that selection of these elements was driven by two radically new biological parameters defining emerging eukaryotic life forms: endosymbiosis with oxygen-metabolizing prokaryotes, and the use of sex in reproduction. Many of the details of these phenomena that would be relevant to senescence were incompletely known to thinkers forty years ago. It is time for a fresh restatement of certain portions of the evolutionary theory in light of recent advances in molecular genetics. This Perspective is an attempt in that direction.

Although occasional cases have been made for processes resembling senescence in prokaryotes (summarized in Ameisen, 1996), senescence is largely an adaptation of eukaryotic life forms. Senescence was present as one of the earliest features of eukaryotic life history, well before the metazoan stage of evolution. The question of whether senescence occurs in the Protista was for many years controversial, but it is now clear that it is in fact quite widespread. In ciliated protozoa such as Paramecia and Tetrahymena, individual cells normally undergo replicative senescence and die in the absence of sex (Bell, 1988). It was noted many years ago that fissile paramecia and other ciliates in culture undergo age-related morphological changes, loss of endocytic function (Smith-Sonneborn and Rodermel, 1976), a decrease in the ability to engage in sex (conjugation), and gradually lose the ability to divide (Maupas, 1889; Pierson, 1938; Dippell, 1955). However, if two ciliates conjugate (Jennings, 1945), or if individual ciliates undergo autogamy (Sonneborn, 1954), the replicative clock is reset and the progeny revert to a "young" appearance.

Orias (Orias, 1986) was among the first to note that events accompanying the degeneration of the parental macronucleus during conjugation are strikingly similar to nuclear degeneration during apoptosis, a key mechanism in programmed cell death in multicellular animals (Ellis, et al., 1991; Ucker, 1991). One of the most highly conserved physiological programs in biology, apoptosis is characterized morphologically by chromatin condensation, cytoplasmic vacuolization and membrane blebbing, and disintegration of the cell into so-called apoptotic bodies that are endocytosed by neighboring cells. Recently, it has been shown that nuclear degeneration in Tetrahymena is accompanied by one of the most distinctive molecular hallmarks of apoptosis, fragmentation of the DNA into nucleosome-size pieces (Davis, et al 1992; Mpoke and Wolfe, 1996).

Thus it would appear that the genetic machinery necessary to execute at least some of the morphological and biochemical elements of programmed cell death were already present in the ciliated protozoa. Apoptotic nuclear destruction has so far only been detected in connection with destruction of the parental macronuclei, which is necessitated by creation of new macronuclei resulting from exchange and recombination of parental micronuclei during sex. The development of separate nuclei in the ciliates would appear to be the first instance in evolution of the segregation of DNA into what would become, in multicellular animals, the germline (micronucleus) and the soma (macronucleus). Since the phenotype of the cells resulting from sexual reproduction is determined by the macronucleus, destruction of the "old" macronucleus is obligatory if the phenotype resulting from the newly generated genotype is to be tested unambiguously against the environment. Cell death in ciliates under suboptimal growth conditions has been proposed to be apoptotic (Christensen, et al., 1995), but whether apoptosis also plays a role in ciliates in the death normally resulting from end-stage replicative senescence in the absence of sex is unclear at present. Nevertheless, the destruction of the macronucleus clearly foreshadows the destruction of the soma in higher eukaryotes at the end of each generation.

Although senescence as a part of the overall life history of the protist slime mold Dyctiostelium discoideum is not well described, programmed cell death is also used by this organism, with many of the cellular features of apoptosis including chromatin condensation, but without the characteristic DNA fragmentation seen in many other cell types (Cornillon, et al, 1994). (DNA fragmentation is also not an absolute requirement of programmed cell death in multicellular organisms: Ellis, et al., 1991; Ucker, 1991) In these protists, programmed cell death is used in a manner that anticipates one of the ways it is used in vertebrates, namely, in morphogenesis. When dictyostelia colonize to form migratory slugs and then fruiting bodies, the stalk cells of the fruiting body become nonviable through programmed cell death (Whittingham and Raper, 1960; Maeda and Taneguchi, 1969; Cornillon, et al., 1994.) Senescence has also been noted in other slime molds (Clark, 1984), and in Plasmodia (Clark and Hakim, 1980), but the underlying mechanisms have not been investigated. In addition to ciliated protozoa and slime molds, senescence and programmed cell death have also been described in the Protista among the kinetoplastid parasites Trypanosoma cruzii (Ameisen, et al., 1995) and Trypanosoma brucei rhodesiense (Welburn, et al., 1996), in the budding ciliate Tokophyra infusionum, and in the green alga Volvox (Kirk, 1997).

Replicative senescence in the Fungi has been well documented (Mortimer and Johnston, 1959; Muller, 1985; Jazwinski, 1990; Kennedy, et al., 1994; Jazwinski, 1996). In filamentous fungi such as Neurospora crassa and Podospora anserina, senescence is determined to a great extent by mitochondrial genetic events modulated in various ways by nuclear genes and adventitious plasmids (reviewed in Jazwinski, 1996). Features of senescence include a gradual decrease of growth rate followed by replicative failure; increased size and granularity of the mycelia; and rearrangements in mitochondrial DNA, followed by its circularization, release and amplification of selected regions (Jamet-Vierny, et al., 1997). It is possible that this circularized DNA, called a-senDNA, encodes a protein with reverse-transcriptase properties (Sainsard-Chanet, et al., 1993). Production of a-senDNA results in major deletions in mitochondrial DNA and gradual mitochondrial degeneration, possibly leading to increased spillage of reactive oxidative species.

Lifespan in yeast, which is defined in terms of mean number of cell divisions rather than calendrically, is genetically determined, in that both maximal and average replicative lifespan can be seen, under laboratory conditions, to be fixed, heritable characteristics of each species or strain. Maximal replicative lifespan is also independent of the environment, e.g., of growth conditions (Muller, et al., 1980). The senescent phenotype is dominant in yeast cells: when young cells are mated with older cells, the resulting zygote reflects the age of the older, rather than the younger, parent (Muller, 1985; Egilmez and Jazwinski, 1989). A detailed genetic analysis of the factors responsible for the dominance of senescence in yeast was one of the most important elements in unraveling the regulation of the eukaryotic cell cycle (Hartwell, et al., 1974; Pringle and Hartwell, 1981; Jazwinski, et al., 1989; Nasmyth, 1993), and led to our current understanding of replicative senescence as a common feature of all dividing eukaryotic cells. The molecular machinery governing the cell cycle in yeast has been remarkably conserved in evolution through the vertebrates (Gautier, et al, 1988; Lee and Nurse, 1988; see also vol. 56 of the Cold Spring Harbor Symposia on Quantitative Biology, 1991, for a comprehensive review.)

In yeast species such as Saccharomyces cerevisiae, daughter cells are produced by budding, and embark on a lifespan characterized by a limited number of cell divisions, usually centering on 20-30 generations (Johnston, 1966; Muller, et al., 1980). The budding process is asymmetric, the mother cell being distinguishable physically from the daughter by the presence of budding scars, and biologically by a reduced proliferative potential. As cells proceed through their life cycle, they also undergo characteristic alterations in cell size and shape, accumulate intracellular granules, and display a gradually increasing cell cycle time. Toward the end of their replicative lifespan, they lose the ability to engage in sexual reproduction (Muller, 1985). The last few cell divisions require a very long time to complete; the cell finally ceases division and eventually dies.

How yeast cells die after end-stage replicative senescence is not clear; yeast cultures always contain a mixture of cell stages, and very old cells are a minute portion of the population. Thus it has not been possible to determine directly whether cells dying at the end of replicative senescence are undergoing apoptosis. However, it has been shown that mammalian bax and bak genes, which induce apoptosis in mammalian cells, are also able to induce cell death in both fission and budding yeasts, and that this death can be blocked by the bcl-2 and bcl-x(L) genes, which block apoptosis in mammalian cells (Greenhalf, et al., 1996; Reed, et al, 1996; Tao, et al., 1997; Jurgensmeier, et al., 1997; Ink, et al., 1997). The bax and bak genes trigger a multicomponent apoptotic pathway in mammalian cells, and Bcl-2 and Bcl-x(L) gene products are thought to block near the end of this pathway. Intriguingly, functional mitochondria are required both for the completion of programmed cell death following replicative senescence, and for bcl-2 rescue, in bax-transformed yeast cells (Greenhalf, et al, 1996).

Bax- and Bak-induced cell death in yeast is accompanied by most of the morphological changes accompanying apoptosis in mammalian cells, such as cytoplasmic vacuolization chromatin condensation, and DNA fragmentation (Ink, et al, 1997). Thus the machinery necessary to execute these cellular changes must be present in yeast. The ceramide-mediated anti-proliferative pathway used to shunt replicatively senescent mammalian cells into a terminal G₁ stage is also present in S. cerevisiae (Nickels and Broach, 1996). It has been suggested that the Bax and Bak proteins may be pore formers, in which case Bax and Bak could conceivably damage yeast cells in the absence of an interaction with intracellular pathways in these cells (Muchmore, et al., 1996). However, the fact that bcl-2, which acts at the mitochondrial membrane and near the end of the sequence of events leading to apoptosis, blocks the lethal effects of the bax and bak genes argues against this possibility. Moreover, human bcl-2 can inhibit programmed cell death induced by other signals in yeast (Longo, et al., 1997). Overall, these data strongly suggest that the programmed cell death pathway manifest in ciliates and Dictyostelium has been conserved in yeast, and is the likely precursor of the pathway found in mammalian cells. The fact that components of the mammalian apoptotic pathway can be used to manipulate cell death in yeast implies a substantial degree of molecular homology between the two systems. However, as with the protists, it is not possible at present to directly link apoptotic cell death and the death that follows replicative senescence in yeast.

Senescence in the nematode worm Caenorhabditis elegans is perhaps one of the most intensely studied aging patterns at the molecular level among early animal life forms. The basic features of C. elegans as an experimental system have been summarized in a comprehensive collection of papers (Wood, 1988). Various strains of C. elegans have average adult lifespans of about two weeks, depending on growth conditions. Senescence in the adult worm is manifested by reduced motility, reduced feeding, and the accumulation of granules containing a lipofuscin-like material. Replicative senescence plays no role in aging of the adult form, since none of the cells in the mature worm divide. However, apoptotic cell death carries forward into C. elegans, where it is used largely in embryonal development. A total of 1,090 cells are generated in the course of producing an adult worm, and 131 of these die at key developmental stages via apoptotic cell death (Sulston, et al., 1983).

The genes underlying programmed cell death in C. elegans have been defined in considerable detail (reviewed in Hengartner and Horvitz, 1994; Hengartner, 1997; Driscoll, 1996). At least a dozen genes have been described that affect this process. The C. elegans genes ced-3 and ced-4 induce programmed cell death. The Ced-3 protein is an enzyme, homologous with the mammalian protein ICE (interleukin-1b converting enzyme), which is also involved in the induction of apoptosis (Yuan, et al., 1993). The nematode ced-3 gene when expressed in mammalian cells can induce apoptotic cell death in vitro (Miura, et al., 1993). Ced-4 is not homologous to any other known protein. The ced-9 gene opposes the effects of ced-3 and ced-4 in cells not fated to die during embryogenesis. The Ced-9 protein is a substrate for Ced-3, and a portion of it is structurally and functionally homologous with the mammalian Bcl-2 protein, which also blocks apoptosis (Hengartner and Horvitz, 1994; Xue and Horvitz, 1997). The human bcl-2 gene can block apoptosis in C. elegans (Vaux, et al., 1992), underscoring the homology of the bcl-2 and ced-9 genes, and the conservation of the basic apoptotic pathway between vertebrates and nematodes. Mammals and C. elegans share at least one other homologous cell death-repressing gene called dad-l (Sugimoto, et al., 1994). Essentially all of the key components of apoptotic programmed cell death as it occurs in vertebrates were already in place at the time of of divergence of the lines leading to vertebrates and C. elegans approximately 600 million years ago (Driscoll, 1996).

In vertebrates, apoptotic programmed cell death is used extensively for morphogenetic events during embryonic development. Replicative senescence is used primarily to shunt microbially or oncologically transformed cells - cells that are dividing in an unscheduled, unregulated fashion from the host’s point of view - into a non-replicative state. Unscheduled cell division is probably synonymous with DNA damage, and/or telomeric degeneration (Kruk, et al., 1994; Vaziri and Benchimol, 1996; Allsop, 1996). In some cases, as will be discussed below, cells may be directed to an apoptotic pathway rather than benign senescence.

Replicative senescence is manifest in vertebrate aging; virtually all vertebrate cells that replicate as a normal part of their life history have been shown to have a limited replicative lifespan in vitro (Stanulis-Praeger, 1987; Cristofalo and Pignolo, 1993; Campisi, et al., 1996) and in vivo (Krohn, 1962; Daniel, et al., 1968; Daniel, 1972). Senescing fibroblasts undergo many of the same morphological changes as yeast cells, and senescence has been shown to be dominant by cell fusion studies. There is considerable correlative evidence linking replicative senescence of cells with the normal aging process in vertebrate organisms (reviewed in Campisi, et al., 1996). However, it seems unlikely that replicative senescence is itself a primary cause of aging.

It is unclear what happens to fibroblasts that become replicatively senescent. Nondividing cells can be maintained in a viable state for a considerable time in vitro (Matsumura, et al., 1979; Pignolo, et al., 1994; Wang, et al., 1994). The cells, by morphological and some biochemical criteria, are clearly in an advanced senescent state. It is unlikely that they are in any sense immortal, but like ciliates and yeast, no one has yet observed them as they actually die, so we do not know whether they die by apoptosis or necrosis. A clarification of the nature of death experienced by cells - at all stages of evolution - resulting from replicative senescence would contribute a great deal to our understanding of the relationship of replicative senescence and programmed cell death. Although these two processes arose at about the same time in evolution, their exact relationship remains a mystery.

The origin and evolution of oxidative mechanisms of cellular senescence.

Programmed cell death and replicative senescence are very likely caused by what is almost certainly the major senescence-inducing challenge faced by eukaryotes immediately upon their evolutionary emergence: oxidative damage to biological macromolecules. According to the oxidative stress hypothesis of aging, cellular senescence is proposed to result from cumulative irreversible damage caused by toxic oxygen intermediates (Ames, et al., 1993; Sohal, et al., 1993; Edgington, 1994; Yu and Yang, 1996; Berlett and Stadtman, 1997). Aging of eukaryotic organisms is seen as the cumulative sum of unrepaired oxidative injuries at the cellular level, triggering in multicellular organisms idiopathic disease which renders the organism more susceptible to all forms of accidental death, and which may in itself be fatal in longterm survivors (e.g., cancer or cardiovascular disease.) The net result in terms of damage to organisms over time is presumed to represent a balance between the rate of production of reactive oxygen species, and the effectiveness of either their neutralization or repair of the cellular damage they cause, particularly to DNA. A decrease in repair of damage to DNA is one of the concomitants of aging in a wide range of species (Weraarchakul, et al., 1989; Wei, et al., 1993; Shikenaga, et al., 1994; Barnett and King, 1995; Gilchrest and Bohr, 1997; Gaubatz and Tan, 1997). One of the attractions of the oxidative damage hypothesis is that it suggests how a limited number of housekeeping genes, involved either in the generation or repair of oxidative damage, could cause the very broad array of phenotypes associated with aging. It is possible that the recently described gene underlying Werner’s syndrome, Wrn (Yu, et al., 1996; Bennett, et al., 1997; Gray, et al., 1997), which is a DNA helicase, could be an example of such a gene. Interestingly, mutants of the yeast homolog of Wrn, Sgs-1, also cause accelerated aging.

Oxidative damage is caused by reactive oxygen species (hereinafter referred to as oxygen radicals) generated by cells through several pathways. The most widespread source in all eukaryotic cells is the mitochondrion, which releases small but constant amounts of oxygen radicals during the reduction of oxygen to water through the electron transport chain. Oxygen radicals are also produced during detoxification processes such as those mediated by cytochrome p450, during the catabolism of excess lipids in peroxisomes, and, in multicellular animals, by phagocytes during clearance of microbes and dead self cells. The most damaging of the oxygen radicals by far, particularly to DNA, is the hydroxyl radical. Eukaryotic cells have evolved several endogenous defenses against hydroxyl radicals, including superoxide dismutase (SOD) and the various catalase enzymes, and DNA repair enzymes. The level of these activities correlates well with senescence and lifespan (Tolmasoff, et al., 1980; Weirich-Schweiger, et al., 1994). Substances taken in through the diet, such as vitamins A, C and E, can also be important in coping with oxidative damage.

Without doubt, oxidative damage is a major byproduct of the incorporation of oxygen-processing bacteria into eukaryotic cells (Margulis, 1993; Frade and Michelidis, 1997), which may well have been one of the founding events in the emergence of eukaryotes. Unfortunately, there is little reliable information on either the generation or repair of oxidative damage in the earliest eukaryotes (e.g. the Protista). Transient UV-induced DNA damage has been reported to increase clonal lifespan in Paramecia, presumably through induction of enhanced DNA repair (Smith-Sonneborn, 1979), which could be an important response to oxidative damage as well. However, the generation of oxygen radicals in yeast, and the systems that control damage by these radicals, have been extensively described (reviewed in Moradas-Ferreira, et al., 1997). The principle sources of oxygen radicals in yeast are the mitochondria and peroxisomes (Gralla, 1997). Yeast have both a peroxisomal and a cytoplasmic catalase; the latter is highly homologous to catalases from vertebrates (Nakagawa, et al., 1995). Yeast also have a Cu/Zn (SOD1; cytoplasmic) and a Mn (SOD2; mitochondrial) superoxide dismutase, in addition to a peroxidase and a glutathione reductase. The genes for both SODs have been cloned and sequenced, and are more than 50 percent homologous to mammalian SODs (Lerch and Schenk, 1985; Beyer, et al., 1991). Together with a metalothionein, a thioredoxin pathway, and heat shock proteins, yeast are well equipped to scavenge the immediate products of oxygen radical production. Yeast also have a system for repairing oxidative damage in cells, including oxidative damage to DNA. One of the genes involved in this repair, called ogg1, is more than 40 percent identical with a homologous repair gene in humans (Arai, et al., 1997; Rosenquist, et al., 1997).

An interaction between oxidative damage/repair pathways and apoptotic programmed cell death in yeast is suggested by recent experiments showing that the human bcl-2 gene expressed in yeast cells lacking SOD1, which cannot survive in the stationary phase in an oxygen atmosphere, protects the cells from oxidative damage and prolongs their survival (Longo, et al., 1997). These results provide further evidence that apoptotic cell death is a normal part of the life cycle of yeast, and very likely the proximal cause of the cell death experienced at the end of replicative senescence.

It is quite clear at the molecular level that yeast already have in place both the major causes of oxidative damage (oxygen radicals leaking from mitochondria and peroxisomes), and antioxidant defenses that are the direct genetic precursors of those found in higher organisms. The notion that these same defenses were carried forward in eukaryotic evolution is supported by findings in C. elegans. Lifespan in C. elegans correlates very well with effectiveness of antioxidant defenses (Larsen, et al., 1993), and the ability to deal with oxidative stress declines with age (Darr and Fridovitch, 1995). The genes for the C. elegans Cu/Zn and Mn SODs have been cloned (Larsen, et al, 1993; Giglio, et al, 1994a,b), and can be seen to share extensive sequence homology with the corresponding genes in both yeast and humans. There is also a set of so-called "clock" genes in C. elegans, which can have major effects on lifespan, and which appear to function by regulating the overall level of metabolic activity in cells (Wong, et al., 1995; Hekimi, et al., 1995; Lakowski and Hekimi, 1996; Hengartner, 1997). The first of these to be cloned and sequenced (clk-1) turns out to be highly conserved in yeast and humans. Although its exact function in higher eukaryotes is not known, it may well function as a regulator of stress responses (Ewbank, et al., 1997).

Mutations in the age-1 gene in C. elegans are particularly interesting. Some alleles of age-1 can confer a near doubling of lifespan, and this is accompanied by reduced fecundity early in life (Friedman and Johnson, 1988), as predicted by Williams’ theory (Williams, 1957). Worms carrying these variants show increased resistance to both oxidative and thermal stress. Although the product of the age-1 gene has not yet been identified, the age-increasing alleles of this gene are accompanied by a decrease in the accumulation of DNA damage, an increase in the ability to repair UV-induced DNA alterations, decreased mitochondrial damage, and an increase in intracellular levels of antioxidant enzymes such as superoxide dismutase and catalase (Larsen, 1993; Melov, et al., 1995; Lithgow, et al., 1995). While often cited as a classic example of a gene selected through antagonistic pleiotropy, it should be noted that age-1 is almost certainly an example not of a gene that encodes a senescence effector protein, but rather a gene that regulates the expression of other senescence effector mechanisms. The selected (wild-type) alleles all allow senescence to proceed faster than the so-called mutant alleles.

In vertebrates, genes such as p53 play a key role in mediating the response to oxidative damage. The p53 protein is a transcriptional activator, and plays a key role in the growth arrest of proliferating cells either at the end of their normal replicative lifespan, or in instances where they become abnormally transformed (Bond, et al., 1996). In the case of cells with severely damaged DNA (including terminally degraded telomeres), p53 may shunt the cell directly into apoptotic cell death (Oren, 1994; Vaziri and Benchimol, 1996). As with the effect of bcl-2 on replicative senescence in yeast (Longo, et al., 1997), this provides another direct link between replicative senescence and apoptotic programmed cell death. It has recently been proposed that p53 may act by enhancing the production of reactive oxygen species, thus hastening the demise of cells with damaged DNA (Johnson, et al., 1996; Polyak, et al., 1997), probably through accelerated destruction of the mitochondrion (Vayssiere, et al., 1994; Polyak, et al., 1996). p53 also is known to induce the death-inducing Bax protein (Han, et al., 1996).

A direct molecular homolog of p53 in early eukaryotes has not yet been identified, but there is abundant evidence that the molecular pathway triggered by p53 is present in yeast. Strong growth inhibition is induced in S. pombe by human p53, and mutants which compromise p53 function in humans block its function in yeast (Bischoff, et al., 1992; Nigro, et al., 1992; Casso and Beach, 1996). The ability of p53 to induce transcription in S. cerevisiae was shown to depend on the yeast gene PAK 1 (Thiagalingam, et al., 1995). An S. cerevisiae mutant that requires p53 for growth was examined, and a yeast protein (Rft1-1) was identified with which the p53 protein interacts (Koerte, et al., 1995). The RAD9 family of yeast cell cycle checkpoint proteins bear a motif common to vertebrate proteins that interact with p53 (Bok, et al., 1997). Thus it is almost certain that the pathway by which p53 directs cells toward apoptotic cell death in vertebrates was already present in yeast. Related genes in C. elegans are less evident, probably because of the paucity of dividing cells in adult worms. However, a homolog of the mammalian gene EI24, a p53-dependent gene involved in the response of mammalian cells to oxidative damage, has been identified (Lehar, et al., 1996).

From the foregoing review, a number of points emerge that bear in an important way on how we think about the evolution of senescence.

1) It is clear that senescence, in the form of replicative senescence, appeared at the earliest stages of eukaryotic evolution. Like all organismal senescence, replicative senescence in unicellular eukaryotic organisms has death as its obligatory endpoint; we see this in both the Protista and the Fungi. Replicative senescence in these early eukaryotes shares most of the morphological features seen in replicative senescence of dividing cells in vertebrates. As far as can be discerned from the data available at present, the molecular basis of replicative senescence is remarkably conserved as well. For example, the mechanisms underlying the eukaryotic cell cycle, which controls replicative senescence, have changed little between yeast and humans, and direct molecular homologies in these mechanisms are abundantly apparent between the two systems. Even more impressive, perhaps, is the fact that mutations in the yeast homolog of the Werner’s syndrome gene cause an accelerated aging phenotype in yeast with similarities at the cellular level to that seen in Werner’s syndrome itself (Sinclair, et al., 1997). For unicellular organisms, the genes underlying replicative senescence certainly qualify as senescence effector genes.

2) Apoptotic nuclear destruction, which is the critical element of programmed cell death, seems to have appeared at about the same time as replicative senescence. What cannot be determined at present is whether the programmed death that comes to unicellular eukaryotes at the end of replicative senescence is based on the same type of apoptotic destruction seen in the macronucleus of ciliates and in other protists. There is strong indirect evidence for the presence of apoptotic programmed cell death in many protists and in yeast, but direct experimental evidence for a link between replicative senescence and apoptosis is admittedly lacking. Nevertheless, as with replicative senescence itself, the genes involved in apoptotic cell death appear to be remarkably conserved throughout the eukaryotes, and for unicellular eukaryotes qualify as senescence effector genes.

3) While senescence as a biological process is clearly established in the earliest single-cell eukaryotes, replicative senescence per se is almost certainly not a primary cause of senescence in most multicellular animals. Its importance lies in establishing the existence of senescence leading to cell death as a process in the earliest eukaryotes. The best evidence at present strongly suggests that oxidative damage to macromolecules, particularly nuclear and/or mitochondrial DNA, caused largely by spillage of reactive oxygen species from mitochondria and other cell organelles, together with a declining ability to repair this damage after reproduction has commenced, is the dominant internal cause of cellular senescence in virtually all multicellular animals, which in turn is the major cause of organismal senescence. Senescence in its simplest terms represents an interplay between senescence effector genes and senescence repressor genes.

It is likely that unrepaired oxidative damage is itself the primary cause of replicative senescence in many dividing eukaryotic cells, including the earliest protists. The relationship of telomeric shortening (Harley, et al., 1992; Vaziri and Benchimol, 1996; Chiu and Harley, 1997), nucleolar rDNA degradation (Guarente, 1997; Sinclair, et al., 1997) and heterochromatin dispersal (Villeponteau, 1997) to oxidative damage in dividing cells remains to be elucidated. It is possible that one or more of these may be an independent senescence mechanism, but again, the genes involved are stringently conserved throughout all of eukaryotic evolution (see e.g. Sinclair, et al., 1997).

The senescence effector genes underlying the generation of oxidative damage, the attenuation of senescent damage by antioxidant enzymes, and the ability to repair senescence-related damage, are all well established in the earliest eukaryotes, and have been maintained with relatively minor change at the molecular level throughout eukaryote evolution. The same would be true for potential telomere- and heterochromatin-based senescence mechanisms, should these prove to be independent of oxidative damage mechanisms. The genes for antioxidant enzymes and DNA repair molecules certainly qualify as senescence resistor genes, as would heat shock protein genes. There has in the past been a general reluctance to accept the possibility that a limited number of genes and mechanisms could underlie organismal senescence. But if it is accepted that organismal senescence is entirely understandable in terms of cellular senescence, then it must be admitted that at the cellular level, virtually everything we understand about senescence can be accounted for by the oxidative stress hypothesis (see e.g. Nohl, et al., 1997).

It is against this current understanding of the molecular basis of senescence and cell death that those portions of the evolutionary theory of senescence dealing with the origin and evolution of the genes causing senescence must be reevaluated. It is apparent from reading the works of Medawar (Medawar, 1945; 1952) and Williams (Williams, 1957) that they envisioned senescence in terms of mechanisms and processes that accumulated more or less independently in many different species, and over very long periods of time. Moreover, Williams clearly thought that senescence in each physiological system in the body would be governed by different mechanisms. He stated outright that he considered the idea of a limited number of underlying mechanisms "a logical impossibility." (Williams, 1957, p. 407.) He gave as one example of a senescence effector gene, that might have been selected through antagonistic pleiotropy, a hypothetical gene involved in calcification of bone. This gene would play a positive role in the fetal development of an individual, hence contributing to the reproductive efficiency of that individual, but later in life the same gene could cause calcification of arteries, thus promoting senescence. Williams strongly implied that these would be the types of genes which, in toto, would account for senescence. Medawar had previously identified the gene presumed to underlie Huntington’s disease as a senescence-causing gene that could have accumulated randomly in the human genome (Medawar, 1952).

Alex Comfort, one of the dominant figures in thinking about senescence in the early decades of the second half of this century, stated that "Senescence...has arisen by convergent evolution in a number of groups. It is not an ‘inherent property’ of the Metazoa, but one which they have on several occasions acquired....In this respect the senescence of insects and man is a comparable process only in the sense that the eyes of these organisms are comparable structures....Such a concept...excludes general physiopathological theories of its [senescence’s] causation." (Comfort, 1954, p. 319.)

The model for the evolution of senescence these views lead to is one of gradual, random accumulation of allelic variants over time, resulting in idiosyncratic collections of senescence effector genes in each species, depending on the physiology of its members and to some extent on their environment. There would indeed be no reason under such a scenario to expect that senescence in humans and insects would be in any way similar. What would life-limiting processes in Drosophila, for example, have in common with Huntington’s disease, cardiovascular disease, Alzheimer’s disease or ALS in humans? But in fact it is clear that all four of these senescence-related diseases in humans (Ames, et al., 1993; Hensley, et al., 1996; Browne, et al., 1997; Jenner, 1996; Rabizadeh, et al., 1995), as well as lifespan in Drosophila (Orr and Sohal, 1994), are correlated with oxidative damage.

What a careful reading of the literature of the past two decades concerning the molecular basis of senescence makes abundantly clear is that far from being caused by random, idiosyncratic collections of genes in each species, senescence effector genes as well as senescence resistor genes in eukaryotes are in fact remarkably constant across all species, and are organized into a highly limited number of senescence mechanisms (Leist and Nicotera, 1997). That the number of genes involved is limited is suggested by the fact that mutations in single genes in both animals (e.g., age-1 in C. elegans; lag-1 in yeast) and humans (any of the single-gene progerias, e.g. Werners syndrome) can trigger major portions of the aging phenotype. Antagonistic pleiotropy could conceivably account for the appearance of alleles that increase an individual’s sensitivity to basic mechanisms of senescence already in place, principally oxidative damage; the e4 allele of the ApoE gene would be a particularly good example because of its involvement in a range of senescent pathologies. The allele of the DNA helicase causing Werner’s syndrome, if it turns out to be involved in DNA repair, could be another example. But oxidative damage as an effector mechanism of senescence is extraordinarily potent, and must have been so from the very beginning; it probably did not require much further evolutionary refinement after the emergence of eukaryotes. Even now, life forms that die in a matter of weeks or months die from the very same underlying mechanisms as animals living fifty or a hundred years. Thus it is unlikely that much evolutionary energy has been spent generating senescence effector genes. The same would seem to be true of senescence resistor genes. A more important task in evolution would appear to have been the accrual, where it makes sense in terms of the life history of individual organisms (Kirkwood, 1981; Partridge and Barton, 1993), of senescence regulatory genes that accelerate or delay the onset of senescence, or increase or decrease its effects. Regulatory genes that delay or mitigate senescence could evolve through normal processes of natural selection, and could account for the more or less steady tendency in evolution toward longer lifespans. The evolutionary development of senescence regulatory genes that accelerate or retard the onset of senescence could very well be accounted for by strategies such as antagonistic pleiotropy. The target of senescence regulatory mechanisms would likely be senescence resistor genes such as those for antioxidant enzymes, heat shock proteins or DNA repair systems (see e.g. Bolzan, et. al, 1997).

Thus what has occurred over evolutionary time is likely not the accretion of genes or new alleles responsible for effecting or repairing the damage associated with senescence, but rather of genes controlling the exposure of eukaryotic organisms to greater or lesser risk, at earlier or later times in their life histories, from a relatively limited and evolutionary ancient number of senescence effector mechanisms. Breaking the genetic basis of senescence down into genes that effect or resist senescence, and senescence regulatory genes that control its onset and rate, was in fact implicit in the proposals of Medawar in the 1950s and even Haldane in the 1940s, but this concept was overlooked by others trying to understand how senescence effector genes could have evolved in the first place. This approach, however, allows us to divide our consideration of the evolution of senescence into two parts: a consideration of the biological needs and selective pressures driving the emergence and fixation of senescence in the very earliest stages of eukaryote evolution on the one hand, and on the other hand a consideration of the biological needs and selective pressures driving each species to control the timing of onset and the rate of senescence in a particular manner. It is in connection with this latter task, the evolutionary development of senescence regulatory mechanisms, that strategies such as antagonistic pleiotropy are most likely to have played a role, rather than in the initial selection of senescence effector genes.

The evolutionary origin of senescence as a process is almost certainly congruent with two events that appear to have happened very close in time, and very early in eukaryote evolution: the incorporation of oxygen-metabolizing prokaryotic cells as a means of dealing with the oxygen crisis, and the incorporation of sex into reproduction. Various scenarios for the inclusion of prokaryotic cells into early eukaryotes have been described (Taylor, 1974; Margulis, 1993; Margulis, 1996). The key role of the prokaryote-derived mitochondria in apoptotic cell death has been noted by numerous authors (Jazwinski, 1996; Liu, et al., 1996; Susin, et al., 1996; Marchetti, et al., 1996; Petit, et al., 1996; Zamzami, et al., 1996; Jamet-Vierny, et al., 1997; Yakes and Van Houten, 1997). A requirement for the presence of mitochondrial DNA for normal apoptosis has also been noted (Greenhalf, et al., 1997), and mitochondria are also key in defenses against apoptosis (Longo, et al., 1997). Coherent models for induction of apoptosis by mitochondria under conditions of stress have been formulated, based on host control of mitochondria-associated bacterial porins, and loss of this control under conditions of cellular stress (Frade and Michaelidis, 1997), or on damage to mitochondrial DNA (Miquel, 1998) which, unlike nuclear DNA, is not repairable (Ozawa, 1995).

The incorporation of oxygen-metabolizing bacteria into early eukaryotic cells in a sense represents a case of antagonistic pleiotropy: the reproductive advantage of using oxygen as a source of energy may well have outweighed its negative impact in terms of senescence and accelerated death of the post-reproductive individual. The prokaryotes themselves had already developed defenses against oxidative damage such as catalases and superoxide dismutases, and doubtless some of these were brought into the new eukaryotic hosts as well, although some of the eukaryotic antioxidant defenses appear to have arisen independently (Finch, 1990, p.570; Kroll, et al., 1995). Selective pressures favoring the development of such additional defenses are self-evident. Thus the basic mechanisms for programmed cell death - a set of senescence effector genes along with senescence resisting genes to forestall senescence until reproduction commenced - may have been acquired by eukaryotes together with the bacteria they imported during pivotal endosymbiotic events.

But if these elements were all largely present in prokaryotes, why then do we not see senescence in prokaryotes? A case has been made that some forms of what could be regarded as programmed death can be found among the bacteria (Amiesen, 1996), although whether these mechanisms appeared before emergence of the eukaryotes is unclear, and at any rate they appear to bear little resemblance to programmed death in eukaryotes. The case can also been made that sex may have been a driving force in the evolution of senescence (Bell, 1982; Clark, 1996), possibly through its impact on DNA correction and repair (Michod, 1993; Long and Michod, 1995), a major corollary of oxidative stress. In the case of the ciliates, apoptotic nuclear destruction serves the important function of removal of genetically "old" macronuclear DNA, so that the new micronuclear genetic combinations produced by sex can be used to generate new macronuclear DNA, and thus unambiguously determine the phenotype of the new daughter cells. It is difficult to imagine the evolutionary continuation of the ciliates and their derivative life forms if this problem had not been resolved. In these organisms, positive selection of genetic mechanisms facilitating removal of macronuclear DNA would clearly further the immediate interests of the newly produced micronuclear DNA; we would not need to resort to hypothetical group selection strategies to explain their continued presence in individuals. In this sense, a death mechanism - destruction of the macronucleus - could be viewed as having been positively selected in evolution.

We still do not know exactly why sex evolved, but it is clear that it evolved in a number of ways in different unicellular eukaryotes. Not all such organisms dealt with the problem of ever-increasing size by developing a "somatic" nucleus that would have to be destroyed each time they had sex. But the condition of having somatic and germline functions combined in a single cell placed extraordinary restrictions on early eukaryotes, and a close examination of how other unicellular eukaryotes dealt with these restrictions may well reveal further correlations of sex with senescence.

In terms of the second important component of the biological bases and selective pressures involved in the evolution of senescence, namely how the timing and rate of its expression are controlled, the evolutionary theory in its current form has been highly successful in describing what we actually see. Williams’ predictions of how fecundity and accidental death rates would affect the evolution of senescence in individual species, of the correlation between rapid individual development and early onset of senescence, and especially of the tight linkage between reproductive maturity and the onset of senescence in all species, (Williams, 1957), have all largely been validated by experiments based on population genetics (reviewed in Comfort, 1979; Finch, 1990; Rose, 1991; see also Rose, 1984; Luckinbill and Clare, 1985; Reznick, 1997). Williams was for the most part silent on the nature of the genes that might control the timing of senescence, but given the tight correlation of reproductive maturity and the onset of senescence there can be little doubt that in many species reproductive hormones play a major role (see for example Finch, 1990, chapter 10; Finch and Rose, 1995; Bolzan, et al., 1997). Genes such as age-1 in C. elegans would also fall into this category; the molecular identity of this gene will be of great interest to studies of the regulation of senescence. Determining exactly how the timing and force of senescence is controlled through such chemical signals will doubtless shed new light on the evolution of senescence.

The first portion of Williams’ classic 1957 paper contained minor refinements of ideas already put forward in a more forceful way by Medawar, who in turn was expanding notions contributed by Haldane, who was himself pondering evolutionary problems identified even earlier by others (e.g., Bidder, 1932). These ideas, concerning the means by which genes responsible for senescence and ultimately compulsory (programmed) death could have crept into the eukaryotic genome, form a coherent intellectual lineage that should be carved out from our notion of what constitutes the evolutionary theory of aging. Although highly imaginative at the time, these early ideas about the origin of senescence-inducing genes are almost certainly unnecessary. While on occasion genes that cause the underlying cellular damage responsible for senescence may simply accumulate as late-acting harmful alleles, or be selected by something resembling antagonistic pleiotropy, it is highly unlikely such genes account for anything more than a minor portion of senescence effector genes in eukaryotes. On the other hand, it is entirely possible that some of the genes accelerating or retarding the onset and/or intensity of senescence - senescence regulator genes - arose in this fashion. The remaining parts of Williams’ ideas, which underlie the rest of the evolutionary theory of senescence, would be valid regardless of how senescence effector genes were originally selected; they are in no way dependent on antagonistic pleiotropy per se. These ideas have proved extremely valuable in guiding research, have been found to have strong predictive value, and should stand alone in what we refer to as the evolutionary theory of senescence.

ALLSOPP, R. 1996. Models of initiation of replicative senescence by loss of telomeric DNA. Exp. Gerontol. 31:235-240

AMEISEN, J., et al. 1995. Apoptosis in a unicellular eukaryote (Trypanosoma cruzii): implications for the evolutionary origin and role of programmed cell death in the control of cell proliferation, differentiation and survival. Cell Death Differ. 2:285-300

AMIESEN, J. 1996. The origin of programmed cell death. Science 272:1278-1279

AMES, B., et al. 1993. Oxidants, antioxidants and the degenerative diseases of aging. Proceedings of the National Academy of Sciences 90:7915-7922

ARAI, K., et al. 1997. Cloning of a human homolog of the yeast OGG1 gene that is involved in the repair of oxidative DNA damage. Oncogene 14:2857-2861

BARNETT, Y., and C. KING. 1995. An investigation of anti-oxidant status, DNA repair capacity, and mutation as a function of age in humans. Mut. Res. 338:115-128

BELL, G. 1982. The Masterpiece of Nature: The Evolution and Genetics of Sexuality. Univ. of Calif. Press

BELL, G. Sex and Death in Protozoa. 1988. Cambridge University Press, Cambridge.

BENNETT, S., et al. 1997. Mismatch repair in extracts of Werner syndrome cell lines. Cancer Research 57:2956-2960

BERLETT, B. and E. STADTMAN. 1997. Protein oxidation in aging, disease, and oxidative stress. J. Biol. Chem. 272:20313-16

BEYER, W., J. IMLAY and I. FRIDOVITCH. 1991. Superoxide dismutases. Progr. Nucl. Acid Res. Molec. Biol. 40:221-253

BIDDER, G. 1932. Senescence. Brit. Med. J. 115:583-585

BISCHOFF, J., D. CASSO and D. BEACH. 1992. Human p53 inhibits growth in Schizosaccharomyces pombe. Molec. Cell. Biol. 12:1405-1411

BODNAR, A., et al. 1998. Extension of lifespan by introduction of telomerase into normal human cells. Science 279:349-52

BOLZAN, A., et al. 1997. Relationship between pituitary hormones, antioxidant enzymes, and histopathological changes in mammary gland of senescent rats. Exp. Gerontol. 32:297-304

BORK, P., et al. 1997. A superfamily of conserved domains in DNA damage-responsive cell-cycle checkpoint proteins. Faseb J. 11:68-76

BROWNE, S., et al. 1997. Oxidative damage and metabolic dysfunction in Huntington's disease: selective vulnerability of the basal ganglia. Annals of Neurology, 1997 May, 41:646-53.

CAMPISI, J., G. DIMRI and E. HARA. 1996. Control of replicative senescence. In, Handbook of Aging, Fourth Edition. Academic Press, New York. pp 121-149

CASSO, D. and D. BEACH. 1996. A mutation in a thioredoxin reductase homolog suppresses p53-induced growth inhibition in the fission yeast Schizosaccharomyces pombe. Molec. Gen. Genet. 252:518-529

CHARLESWORTH, B. 1994. Evolutionary mechanisms of senescence. In, M. Rose and C. Finch, eds., Genetics and Evolution of Aging. Kluwer Academic Publishers, Amsterdam,1994. pp. 13-21

CHIU, C. and C. HARLEY. 1997. Replicative senescence and cell immortality: the role of telomeres and telomerase. Proc. Soc. Exp. Biol. Med. 214:99

CHRISTENSEN, S., et al. 1995. Mechanisms controlling death, survival and proliferation in a model unicellular eukaryote, Tetrahymena thermophila. Cell Death Diff. 2:301-308

CLARK, J. 1984. Lifespan and senescence in six slime molds. Mycologia 76:366-368

CLARK, J. and HAKIM, R. 1980. Aging of plasmodial heterokaryons in Didymium iridis. Molec. Gen. Genet. 178:419-422

CLARK, W. 1996. Sex and the origins of death. Oxford University Press, New York.

COMFORT, A. 1954. Biological aspects of senescence. Biological Reviews 29:284-329

CORNILLON, S., et al. 1994. Programmed cell death in Dictyostelium. Journal of Cell Science 107:2691-2704

CRISTOFALO, V., and R. PIGNOLO. Replicative senescence of human fibroblast-like cells in culture. 1993. Physiological Reviews 73:617-638

CURTSINGER, J., P. SERVICE and T. PROUT. 1994. Antagonistic pleiotropy, reversal of dominance, and genetic polymorphism. Amer. Naturalist 144:210-228

CUTLER, R. 1979. Evolution of human longevity: a critical overview. Mech. Age. Devel. 9:337-354

DANIEL, C. 1972. Aging of cells during serial propagation in vivo. Adv. Geront. Res. 4:167-199

DANIEL, C., et al. 1968. The in vivo lifespan of normal and preneoplastic mouse mammary glands: a serial transplantation study. Proc. Nat. Acad. Sci. 61:53-60

DARR, D. and I. FRIDOVICH. 1995. Adaptation to oxidative stress in young, but not old, Caenorhabditis elegans. Free Rad. Biol. Med. 18:195-201

DAVIS, M., et al. 1992. Programmed nuclear death: apoptotic-like degeneration of specific nuclei in conjugating Tetrahymena. Developmental Biology 154:419-432

DIPPELL, R. 1955. Some cytological aspects of aging in variety 4 of Paramecium aurelia. J. Protozoology 2(suppl):7-

DRISCOLL, M. 1996. Cell death in C. elegans: molecular insights into mechanisms conserved between nematodes and mammals. Brain Pathology 6:411-425

EDGINGTON, S. 1994. As we live and breathe: free radicals and aging. Bio/Technology 12:37-40

EGILMEZ, N. et al. 1989. Specific alterations in transcript prevalence during the yeast lifespan. J. Biol. Chem. 264:14312-14317.

EGILMEZ, N. and M. JAZWINSKI. 1989. Evidence for the involvement of a cytoplasmic factor in the aging of the yeast Saccharomyces cerevisiae. J. Bacteriol. 171:37-42.

ELLIS, R., YUAN, J., and HORVITZ, R. H. 1991. Mechanisms and function of cell death. Ann. Rev. Cell Biol. 7:663-698

EWBANK, J., et al. 1997. Structural and functional conservation of the Caenorhabditis elegans timing gene clk-1. Science 275:980

FINCH, C. 1990. Longevity, senescence and the genome. University of Chicago Press, Chicago.

FINCH, C. and M. ROSE. 1995. Hormones and the physiological architecture of life history evolution. The Quarterly Review of Biology 70:1-36

FRADE, J., and T. MICHAELIDIS. 1997. Origin of eukaryotic programmed cell death: a consequence of aerobic metabolism? BioEssays 19:827-832

FRIEDMAN, D. and T. JOHNSON. 1988. A mutation in the age-1 gene in Caenorhabditis elegans lengthens life and reduces hermaphrodite fertility. Genetics 118:75-86

GAUBATZ, J. and B. TAN. 1997. Aging affects the levels of DNA damage in post-mitotic cells. Ann. NY Acad. Sci. 719:97-107

GAUTIER, J., et al. 1988. Purified maturation-promoting factor contains the product of a Xenopus homolog of the fission yeast cell cycle control gene cdc2. Cell 54:433-439

GIGLIO, M., et al. 1994a. The manganese superoxide dismutase gene of Caenorhabditis elegans. Biochem. and Mol. Biol. Intl. 33:41-44

GIGLIO, M., et al. 1994b. The copper/zinc superoxide dismutase gene of Caenorhabditis elegans. Biochem. and Mol. Biol. Intl. 33:37-40

GILCHREST, B., and V. BOHR. 1997. Aging processes, DNA damage, and repair. Faseb J. 11:322-330

GRALLA, E. 1997. Superoxide dismutase: studies in the yeast Saccharomyces cerevesiae. In, Oxidative Stress and the Molecular Biology of Antioxidant defenses. Cold Spring Harbor Press, Cold Spring Harbor, Mew York. pp 495-525

GRAY, M., et al. 1997. The Werner syndrome protein is a helicase. Nature Genetics 17:100

GREENHALF, W. et al. Role of mitochondria and C-terminal membrane anchor of Bcl-2 in Bax-induced growth arrest and mortality in Saccharomyces cerevesiae. FEBS Letters 380:169-175

GUARENTE, L. 1997. Link between aging and the nucleolus. Genes and Devel. 11:2449-2455.

HALDANE, J. 1941. New paths in genetics. Allen and Unwin, London.

HANS, J., et al. 1996. The E1b 19K protein blocks apoptosis by interacting with and inhibiting the p53-inducible and death-promoting Bax protein. Genes Devel. 10:461-477

HARLEY, C., et al. 1992. The telomere hypothesis of cellular aging. Exp.Gerontol. 27:375-82.

HARTWELL, L., et al. 1974. Genetic control of the cell division cycle in yeast. Science 183:46-50

HEKIMI, S., P. BOUTIS and B. LAKOWSKI. 1995. Viable maternal-effect mutations that affect the development of the nematode Caenorhabditis elegans. Genetics 141:1351-1364

HENGARTNER, M. 1997. Genetic control of programmed cell death and aging in the nematode Caenorhabditis elegans. Exp. Gerontol. 32:363-374

HENGARTNER, M. and H. HORVITZ. 1994. The ins and outs of programmed cell death during C. elegans development. Phil. Trans. Roy. Soc. Lond. Series B. 345:243-246

HENSLEY, K., et al. 1996. Reactive oxygen species as causal agents in the neurotoxicity of the Alzheimer’s disease-associated amyloid beta-peptide. Annals of the New York Academy of Science 786:120

INK, B., et al. 1997. Human Bak induces cell death in Schizosaccharomyces pombe with morphological changes similar to those with apoptosis in mammalian cells. Molecular and Cellular Biology 17:2468-2474

JAMET-VIERNY, C., et al. 1997. Contribution of various classes of defective mitochondrial DNA molecules to senescence in Podospora anserina. Curr. Genet. 31:171-178

JAZWINSKI, S. M., N. EGILMEZ and J. CHEN. 1989. Replication control and cellular lifespan. Experimental Gerontology 24:423-428.

JAZWINSKI, S. M. 1990. Aging and senescence of the budding yeast Saccharomyces cerevisiae. Molecular Microbiology 4:337-343.

JAZWINSKI, S. M. 1996. Longevity assurance genes and mitochondrial DNA alterations: yeast and filamentous fungi. Handbook of the Biology of Aging (fourth edition), p. 39 Academic Press, New York.

JENNER, P. 1996. Oxidative stress in Parkinson's disease and other neurodegenerative

disorders. Pathologie Biologie 44:57-64.

JENNINGS, H. 1945. Paramecium bursaria: life history. V. Some relations of external conditions, past or present, to ageing and to mortality of exconjugants, with summary of conclusions on age and death. J. Exp. Zool. 99:15-33

JOHNSON, T., G. LITHGOW and S. MURAKAMI. 1996. Hypothesis: Interventions that increase the response to stress offer the potential for effective life prolongation and increase health. J. Gerontol. 51A:B392-395.

JOHNSON, T., et al. 1996. Reactive oxygen species are downstream mediators of p53-dependent apoptosis. Proc. Natl. Acad. Sci. 93:11848-11852

JOHNSTON, J. 1966. Reproductive capacity and mode of death of yeast cells. Antonie van Leeuwenhoek 32:94-98.

JÜRGENSMEIER, J., et al. 1997. Bax- and Bak-induced cell death in the fission yeast Schizosaccharomyces pombe. Molecular Biology of the Cell 8:325-339

KARAKASHIAN, S., M. LANNERS and M. RUDZINSKA. 1984. Cellular and clonal aging in the suctorian protozoan Tokophyra infusionum. Mech. Age. Devel. 26:217-229

KENNEDY, B., N. AUSTRIACO and L. GUARENTE. 1994. Daughter cells of S. cerevisiae from old mothers display a reduced lifespan. J. Cell Biol. 137:1985

KIRK, D. 1997. Volvox. Cambridge University Press, Cambridge. pp.188-189

KIRKWOOD, T. 1981. Repair and its evolution: survival vs. reproduction. In, Physiological Ecology: An Evolutionary Approach to Resource Use. Blackwell, London, pp 165-189.

KIRKWOOD, T. and M. ROSE. 1991. Evolution of senescence: late survival sacrificed for reproduction. Philosophical Transactions of the Royal Society of London B 332:15-27

KOERTE, A., et al. 1995. Suppression of the yeast mutation rft-1 by human p53. J. Biol. Chem. 270:22556-22564

KROHN, P. 1962. Review lectures on senescence. II. Heterochronic transplantation in the study of aging. Proc. Roy. Soc. London 157:128-147

KROLL, J., et al. 1995. Bacterial Cu/Zn superoxide dismutase: phylogenetically distinct from the eukaryotic enzymes, and not so rare after all. Microbiology 141:2271-2279

KRUK, P., N. RAMPINO and V. BOHR. 1995. DNA damage and repair in telomeres: relation to aging. Proc. Natl. Acad. Sci. 92:258-262

LAKOWSKI, B. and S. HEKIMI. 1996. Determination of lifespan in Caenorhabditis elegans by four clock genes. Science 272:1010-1013

LARSEN, P. 1993. Aging and resistance to oxidative damage in Caenorhabditis elegans. Proceedings of the National Academy of Science 90:89058909

LEE, M. and P. NURSE. 1988. Complementation used to clone a human homologue of the fission yeast cell cycle control gene cdc-2. Nature 327:31-35

LEHAR, S., et al. 1996. Identification and cloning of EI24, a gene induced by p53 in etoposide-treated cells. Oncogene 12:1181-1187

LEIST, M. and P. NICOTERA. 1997. The shape of cell death. Biochem. Biophys. Res. Commun. 236:1-9

LERCH, K. and E. SCHENK. 1985. Primary structure of copper-zinc superoxide dismutase from Neurospora crassa. J. Biol. Chem. 260:9559-9566

LITHGOW, G., et al. 1995. Thermotolerance and extended lifespan conferred by single-gene mutations and induced by thermal stress. Proc. Natl. Acad. Sci. 92:7540-7545

LIU, X., et al. 1996. Induction of apoptotic program in cell-free extracts: requirement for dATP and cytochrome C. Cell 86:147-152

LONG, A. and R. MICHOD. 1995. Origin of sex for error repair. I. Sex, diploidy and haploidy. Theor. Pop. Biol. 47:18-55

LONGO, V., et al. 1997. Human Bcl-2 reverses survival defects in yeast lacking superoxide dismutase and delays death of wild-type yeast. J. Cell Biol. 137:1581-1588

LUCKINBILL, L. and M. CLARE. 1985. Selection for lifespan in Drosophila melanogaster. Heredity 55:9-18

MAEDA, Y. and I. TANEGUCHI. 1969. Cell differentiation and fine structures in the development of the cellular slime molds. Dev. Growth Differ. 11:232-245

MARCHETTI, P., et al. 1996. Mitochondrial permeability transition is a central coordinating event of apoptosis. J. Exp. Med. 184:1155-1160

MARGULIS, L. 1993. Symbiosis in Cell Evolution. Freeman, New York; 2nd ed.

MARGULIS, L. 1996. Archaeal-eubacterial mergers in the origin of Eukaria: Phylogenetic classification of life. Proc. Natl. Acad. Sci. 93:1071-1076

MATSUMURA, T., Z. ZERRUDO, and L. HAYFLICK. 1979. Senescent diploid human cells in culture: survival, DNA synthesis and morphology. J. Gerontol. 34:328-334

MAUPAS, E. 1889. Le rajeunissement karyogamique chez les ciliés. Arch Zool. exp. gen. 7:149-517.

MEDAWAR, P. 1945. Old age and natural death. Modern Quarterly 1:30-33.

MEDAWAR, P. An unsolved problem of biology. An inaugural lecture delivered at University College London, 6 December 1951. H. K. Lewis, London, 1952.

MELOV, S., et al. 1995. Increased frequency of deletions in the mitochondrial genome with age of Caenorhabditis elegans. Nucl. Acid Res. 23:1419-1425

MICHOD, R. 1993. Genetic error, sex and diploidy. J. Hered. 84:360-371

MIQUEL, J. 1998. An update on the oxygen stress-mitochondrila mutation theory of aging: genetic and evolutionary implications. Exp. Gerontol. 33:113-126

MIURA, M., et al. 1993. Induction of apoptosis in fibroblasts by IL-1b-converting enzyme, a mammalian homologue of the C. elegans gene ced-3. Cell 75:653-660

MORADAS-FERREIRA, P., et al. 1997. The molecular defenses against reactive oxygen species in yeast. Molec. Microbiol. 19:651-658.

MORTIMER, R., and J. JOHNSTON. 1959. Lifespan of individual yeast cells. Nature 183:1751-1752

MPOKE, S. and J. WOLFE. 1996. DNA digestion and chromatin condensation during nuclear death in Tetrahymena. Exp. Cell Res. 225:357-365

MUCHMORE, S., et al. 1996. X-ray and NMR structure of human Bcl-X_L, an inhibitor of programmed cell death. Nature 381:335-341

MÜLLER, I., et al. 1980. Calendar lifespan vs. budding lifespan of Saccharomyces cerevisiae. Mechanisms of Ageing and Development 12:47

MÜLLER, I. 1985. Parental age and the lifespan of zygotes of Saccharomyces cerevisiae. J. Microbiol. Serol. 51:1-10

NAKAGAWA, C., N. MUTOH and Y. HAYASHI. 1995. Transcriptional regulation of the catalase gene in the fission yeast Schizosaccharomyces pombe: molecular cloning of the catalase gene and northern blot analyses of the transcript. J. Biochem. 118:109-116

NASMYTH, K. 1993. Control of the yeast cell cycle by the Cdc28 protein kinase. Current Opinions in Cell Biology 5:166-171

NICKLES, J. and J. BROACH. 1996. A ceramide-activated protein phosphatase mediates ceramide-induced G1 arrest in Saccharomyces cerevisiae. Genes Devel. 10:382-394

NIGRO, J., et al. 1992. Human p53 and CDC2H genes combine to inhibit the proliferation of Saccharomyces cerevisiae. Molec. Cell. Biol. 12:1357-1365

NOHL, H., K. STANIEK and L. GILLE. 1997. Imbalance of oxygen activation and energy metabolism as a consequence or mediator of aging. Exp. Gerontol. 32:485-500

OREN, M. 1994. Relationship of p53 to the control of apoptotic cell death. Semin. Canc. Biol. 5:221-227

ORIAS, E. 1986. Ciliate conjugation. In, The Molecular Biology of Ciliated Protozoa. Academic Press, New York. p.45-63

ORR, W. and R. SOHAL.1994. Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263:1128

OZAWA, T. 1995. Mitochondrial DNA mutations associated with aging and degenerative diseases. Exp. Gerontol. 30:269-290

PARTRIDGE, L. and N. BARTON. 1993. Optimality, mutation and the evolution of aging. Nature 362:305-309

PETIT, P., et al. 1996. Mitochondria and programmed cell death. Back to the future. FEBS Lett. 396:7-13

PIERSON, B. 1938. The relation of mortality after endomixis to the prior endomictic interval in Paramecium aurelia. Biol. Bull. (Woods Hole) 74:235-243

PIGNOLO, R., V. CRISTOFALO and M. ROTENBERG. 1993. Senescent WI-38 cells fail to express EPC-1, a gene induced in young cells upon entry into the G₀ state. J. Biol. Chem. 268:8949-8957

PRINGLE, J. and L. HARTWELL. 1981. The Saccharomyces cerevisiae cell cycle. In, The Molecular Biology of the Yeast Saccharomyces: Life Cycle and Inheritance. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York. p. 297

RABIZADEH, S., et al. 1995. Mutations associated with amyotrophic lateral sclerosis convert superoxide dismutase from an antiapoptotic to a proapoptotic gene: studies in yeast and neural cells. Proc. Natl. Acad. Sci. 92:3024-3028

REED, J, et al. 1996. Structure-function analysis of Bcl-2 proteins: Regulators of programmed cell death. Advances in Experimental Medicine and Biology 406:99-112

REZNICK, D. 1997. Life history evolution in guppies (Poecilia reticulata) as a model for studying the evolutionary biology of aging. Experimental Gerontology 32:245-254

ROSE, M. 1991. Evolutionary Biology of Aging. Oxford University Press, New York.

ROSE, M. 1984. Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38:1004-1010 (1984)

ROSENQUIST, T., D. ZHARKOV and A. GROLLMAN. 1997. Cloning and characterization of a mammalian 8-oxoguanine DNA glycosylase. Proc. Natl. Acad. Sci. 94:7429-7424

SHIKENAGA, M., et al. 1994. Oxidative damage and mitochondrial decay in aging. Proceedings of the National Academy of Sciences 91:10771-10775

SINCLAIR, D., K. MILLS and L. GUARENTE. 1997. Accelerated aging and nucleolar fragmentation in yeast Sgs1 mutants. Science 277:1313-1316

SMITH-SONNEBORN, J. 1979. DNA repair and longevity assurance in Paramecium tetraurelia. Science 203:1115-1117

SMITH-SONNEBORN, J. and S. RODERMEL. 1976. Loss of endocytic capacity in aging Paramecium. The importance of cytoplasmic organelles. J. Cell Biol. 71:575-588

STANULIS-PRAEGER, B. 1987. Cellular senescence revisited: a review. Mechanisms of Ageing and Development 38:1-48

SOHAL, R. 1993. The free radical theory of aging: an appraisal of the current status. Aging 5:3-17

SONNEBORN, T. 1954. The relation of autogamy to senescence and rejuvenescence in Paramecium aurelia. J. Protozoology 1:38-53

SUGIMOTO, A., P. FRIESEN, and J. ROTHMAN. 1994. Baculovirus p35 prevents developmentally programmed cell death and rescues a ced-9 mutant in the nematode Caenorhabditis elegans. EMBO J. 13:2023-2028

SULSTON, J., E. SHIERENBERG, J. WHITE and J. THOMSON. 1983. The embryonic cell lineage of the nematode Caenorhabditis elegans. Devel. Biol. 100:64-119

SUSIN, S., et al. 1996. Bcl-2 inhibits the mitochondrial release of an apoptogenic protease. J. Exp. Med. 184:1331-1341

TAO, et al. 1997. Modulation of cell death in yeast by the Bcl-2 family of proteins. J. Biol. Chem. 272:15547-15552

TAYLOR, F. 1974. Implications and extensions of the serial endosymbiosis theory of the origin of eukaryotes. Taxon 23:229-258

THIAGALINGAM, S., K. KINZLER and B. VOGELSTEIN. 1995. PAK-1, a gene that can regulate p53 activity in yeast. Proc. Natl. Acad. Sci. 92:6062-6066

TOLMASOFF, J., T. ONO and R. CUTLER. 1980. Superoxide dismutase: correlation with lifespan and specific metabolic rate in primate species. Proc Natl. Acad. Sci. 77:2777-2781.

UCKER, D. 1991. Death by Suicide: one way to go in mammalian cellular development. New Biol. 3:103-109

VAUX, D., I. WEISSMAN and S. KIM. 1992. Prevention of programmed cell death in Caenorhabditis elegans by human bcl-2. Science 258:1955-1957

VAYSSIERE, J., et al. 1994. Commitment to apoptosis is associated with changes in mitochondrial biogenesis and activity in cell lines conditionally immortalized with simian virus 40. Proc. Natl. Acad. Sci. 91:11752-11756

VAZIRI, H. and S. BENCHIMOL. 1996. From telomere loss to p53 induction and activation of a DNA damage pathway at senescence: the telomere loss/DNA damage model of cell aging. Exp. Gerontol. 31:295-301

VILLEPONTEAU, B. 1997. The heterochromatin loss model of aging. Exp.Gerontol. 32:383-94.

WANG, E., M. LEE and S. PANDEY. 1994. Control of fibroblast senescence and activation of programmed cell death. J. Cell Biochem. 54:432-439

WEI, Q., et al. 1993. DNA repair and aging in basal cell carcinoma: a molecular epidemiology study. Proc. Natl. Acad. Sci. 90:1614-1618.

WEIRICH-SCHWEIGER, H., et al. 1994. Correlation between senescence and DNA repair in cells from young and old individuals and in premature aging syndromes. Mut. Res. 316:37-48

WELBURN, S., et al. 1996. Apoptosis in procyclic Trypanosoma brucei rhodesiense in vitro. Cell Death Differ. 3:229-236

WERAARCHAKUL, N., et al. 1989. The effect of aging and dietary restriction on DNA repair. Experimental Cell Research 181:197-203

WHITTINGHAM, W. and K. RAPER. 1960. Non-viability of stalk cells in Dictyostelium. Proc. Nat. Acad. Sci. 46:642-649

WILLIAMS, G. 1957. Pleiotropy, natural selection and the evolution of senescence. Evolution 11:398-411

WONG A., BOUTIS, P. and S. HEKIMI. 1995. Mutations in the clk-1 gene of Caenorhabditis elegans affect developmental and behavioral timing. Genetics 139:1247-1259

WOOD, W (ed.). 1988. The Nematode Caenorhabditis elegans. Cold Spring Harbor Press, Cold Spring Harbor, New York.

XUE, D. and H. HORVITZ. 1997. Caenorhbditis elegans Ced-9 protein is a bifunctional cell-death inhibitor. Nature 390:305

YAKES, F. and B. VAN HOUTEN. 1997. Mitochondrial DNA damage is more extensive and persists longer than nuclear DNA damage in human cells following oxidative stress. Proc. Natl. Acad. Sci. 94:514-519

YU, C., et al. 1996. Positional cloning of the Werner’s Syndrome gene. Science 272:258-262

YU, B. and R. YANG. 1996. Critical evaluation of the free radical theory of aging. A proposal for the oxidative stress hypothesis. Annals of the New York Academy of Sciences 786:1-11.

YUAN, J., et al. 1993. The C. elegans death gene ced-3 encodes a protein similar to mammalian interleukin-1b-converting enzyme. Cell 74:641-652

ZAMZAMI, N., et al. 1996. Mitochondrial control of nuclear apoptosis. J. Exp. Med. 183:1553-1544