Reach Us +44-1202-068036
Are Internal, Death-Promoting Mechanisms Ever Adaptive? | OMICS International
ISSN: 2329-9002
Journal of Phylogenetics & Evolutionary Biology
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on Medical, Pharma, Engineering, Science, Technology and Business

Are Internal, Death-Promoting Mechanisms Ever Adaptive?

John W Pepper1,2, Deborah E Shelton1, Armin Rashidi1 and Pierre M Durand1,3*
1Department of Ecology and Evolutionary Biology, University of Arizona, Tucson, USA
2National Cancer Institute, Bethesda, Maryland, USA
3Department of Molecular Medicine and Haematology, University of the Witwatersrand, Johannesburg, South Africa
Corresponding Author : Pierre M Durand
Department of Molecular Medicine and Haematology
University of the Witwatersrand, Johannesburg, South Africa
E-mail: [email protected]
Received March 26, 2013; Accepted July 08, 2013; Published July 18, 2013
Citation: Pepper JW, Shelton DE, Rashidi A, Durand PM (2013) Are Internal, Death-Promoting Mechanisms Ever Adaptive? J Phylogen Evolution Biol 1:113. doi: 10.4172/2329-9002.1000113
Copyright: © 2013 Pepper JW, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Related article at
DownloadPubmed DownloadScholar Google

Visit for more related articles at Journal of Phylogenetics & Evolutionary Biology


Natural selection acts primarily on organisms, and the existence of evolved, active, internal mechanisms that cause organismal death would seem paradoxical. However, there is substantial evidence that internal death promoting mechanisms exist and are taxonomically widespread. Where these are argued to be ‘programmed organismal death’ (POD), they require evolutionary explanations. Any such explanation must draw on our understanding of fitness trade-offs and multiple levels of selection in evolution. This review includes two main categories of putative POD: senescence in multicellular-organisms, and programmed cell death in unicellular organisms. The evidence for POD as a genetically controlled phenotype is strong for semelparous and significant but more controversial for iteroparous plants and animals. In multicellular organisms the program frequently (although not always) appears to be the result of fitness trade-offs. Here the death phenotype itself is not adaptive but the fitness related program most likely is. However, in some cases of behavioral suicide, particularly in insects, there are distinct advantages to kin and group level benefits may play a role. In unicells, programmed death is ubiquitous and POD often provides benefits to others. While benefits do not equate with adaptations, they are consistent with it. Here, death may be adaptive at a level other than the individual cell. In other instances of POD in unicells the phenotype (eg autophagy) can be explained as pleiotropy. The overall picture of POD as a natural phenomenon is still emerging, and continued work on diverse lines of evidence is necessary to complete our evolutionary understanding of this apparent paradox. While some questions remain, we conclude that POD is most likely, in some circumstances at least, adaptive.

Programmed death; Adaptation; Unicellular; Organismal
The idea that self-inflicted organismal death could be adaptive sounds, at face value, absurd. An adaptation is a trait that is suitable (apt) for the current circumstances or environmental challenges, and archetypal examples include traits that promote survival. Natural selection is the mechanism that produces adaptations. In describing natural selection, Darwin (1859) emphasized the struggle for survival: “Two canine animals in a time of dearth may be truly said to struggle with each other which shall get food and live [1]. But a plant on the edge of a desert is said to struggle for life against the drought…...”. How could an inherited trait that promotes death, rather than survival, possibly be adaptive?
Four categories encompass the major possible evolutionary explanations for the cause of death of an organism (Table 1). First, death (or an increased probability of death) could inevitably occur despite the efforts or traits of the organism. Second, internal mechanisms that promote death could exist in spite of selective pressure against them. Third, death could occur as a side-effect of a mechanism within the organism that has another function or benefit. Fourth, death could occur because of a mechanism within the organism that evolved explicitly to cause death. This fourth category is the only one in which the mechanism promoting death is an adaptation for promoting death, and cases in this category can only be explained by selection at a hierarchical level other than the organism.
In all categories except the first, we can reasonably expect to see active mechanisms within an organism that promote death. This review was motivated by the observation that diverse organisms apparently have such active, internal death-promoting mechanisms and by the subtle and difficult conceptual issues that understanding the evolution of this kind of trait raises. We use the term programmed organismal death (POD) to refer to organismal death that results directly from an active process that is internally controlled and regulated by the organism (although it may be triggered as a response to an external cue). Several terms from our definition require further explanation.
There are multiple concepts of what constitutes an individual organism for a particular taxon, or a particular biological question [2]. Here, we follow Gould and Lloyd (1999) in using “organism” in the conventional sense to refer to the discrete body of a highly integrated creature [3]. Thus, our idea of POD entails only that the organism in question be a discrete, highly integrated cell or multicellular body. The high integration and functionality that are characteristic of “organisms” (in this sense) are typically, and rightly, understood as a product of organism-level natural selection. However, it is important to note that identifying something as an “organism” in this sense does not imply that the organism level is the only level at which an effective selective process has been occurring or that organism-level selection is the primary cause of all the traits of the organism. For example, we speak of each bacterium within a biofilm as an “organism” (and speak about the active, regulated death of a bacterium as POD) without intending to automatically imply anything about the relative strength of cell-level and biofilm-level selection. Therefore in these and other instances where the cell is the organism (unicellular life forms) the terms POD and PCD (programmed cell death) are used synonymously.
By “death that results directly” and by “internally controlled” we intend to eliminate cases such as predation of an individual due to the “programmed” expression of a sexual ornament as well as cases in which parasites promote the death of a host by changing their behavior. In essence, we consider cases of apparently self-inflicted organismal death.
We consider “active” and “internally controlled and regulated” as sufficient descriptors of “programmed”, although we recognize that this view of the meaning of “programmed” is not ubiquitous [4,5]. We use the word “program” to indicate that the inherited information of a cell can direct responses to the environment; the analogy with computer programs is imperfect, but helpful [6]. Observing that death is “programmed” in this sense simply eliminates the first category of explanations (unavoidable death); further evidence is required to discriminate among the remaining possibilities. Particular aspects of the “programmatic” nature of death may be evidence for or against some of the remaining categories, as summarized in Table 2.
The final term that should be clarified at the outset is “adaptation”, as our central concern is the question of whether death-causing mechanisms are adaptive in some cases. We follow Reeve and Sherman (1993) in defining an adaptation as “a phenotypic variant that results in the highest fitness among a specified set of variants in a given environment” [7]. It is a feature that is apt for the current circumstances, and current utility relative to some field of potential variants is the only criterion. This definition contrasts with others who require various versions of historical criteria to define an adaptation [8-10]. For our current purposes, the distinction is not critical. It is straightforward for those readers who understand “adaptation” to entail historical criteria to apply our summaries of evidence for POD to their own interpretation of “adaptation”, though sufficient evidence to evaluate historical criteria is lacking in many, if not all, cases of putative POD. Ideally, sufficient evidence of adaptation occurs when the putative selective pressure for the phenotype is removed and the trait disappears over time. Some of us and others have acknowledged this [11]; although such an overly stringent criterion is likely impractical for evaluating POD in most organisms including unicells. Controls for such experiments may be technically impossible. Furthermore, failure for a trait to disappear after eliminating putative environmental pressures is also not the final word on the matter and does not exclude conflating phenomena like phenotypic integration, pleiotropy, modularity and weak selection.
Importantly, an affirmative answer to the question of whether a particular feature appears to be an adaptation (sensu [7]) can leave open the question of what it is an adaptation for [12]. Thus, when considering aspects of mechanisms that promote death, some features may simply suggest that POD (i.e., the mechanisms that lead to death) is an adaptation, whereas other features may suggest that POD is an adaptation for a particular function or outcome (e.g. death, regulation of energy allocation). Note again that if death is the function of POD (i.e., if POD is an adaptation for promoting death), then POD must be selected for at a level other than the organism level. Some authors have proposed functional hypotheses for the origin and maintenance of POD without rigorously considering the implied selection process [13]. Others cling dogmatically to the primacy of organism-level selection, even in the face of contradictory evidence [14]. Putative cases of POD (as a natural phenomenon, independent of evolutionary explanations) range from those that are solidly supported by empirical evidence to those that are merely suggested and sometimes controversial. To help establish the adaptive function (or lack thereof) of each putative case of POD, we explicitly consider that levels of selection may have shaped the traits in question.
Justifications of the evidence used to support claims that a particular trait is an adaptation (or is an adaptation for a particular function) are controversial [12]. In-depth treatment of particular lines of evidence or particular cases is beyond the scope of this review. Table 2 summarizes the lines of evidence that we consider with respect to POD and provides references that discuss the subtleties of some types of evidence for selection or adaptation.
This review includes two main categories of putative POD: senescence in macro-organisms and programmed cell death in unicellular microorganisms. Additionally, we briefly consider behavioral-mediated death in macro-organisms. We summarize various forms of evidence (Table 2) that are relevant to discriminating between the categories of death explanations (Table 1). How much evidence is sufficient to definitively assign an example to one of the categories in Table 1 is not straightforward. We generally leave this task to more detailed studies of particular cases, though we do point out categories of cases for which the bulk of the evidence so far points to a particular kind of explanation. Although there is also evidence for senescence of microbial lineages, in the form of loss of replicative potential, we restrict this review to organismal death and exclude reproductive senescence of lineages.
Organismal Senescence as Programmed Death
Clearly it can be adaptive for a large organism to cause the senescence of specific organs or structures, such as leaves on a deciduous plant. Whole-organism senescence leading to death is more puzzling, and explaining it has been a long-standing challenge. If a large organism can build itself from a single cell, why can it not maintain itself once built? Or can organismal death actually be adaptive in some cases, in the sense of increasing Darwinian fitness?
Evidence from Semelparous organisms
The strongest appearance of POD in organismal senescence is found in semelparous species, which reproduce only once during the life cycle. A common pattern in these species is that individuals appear healthy up to the time of reproduction, then decline rapidly and dramatically afterward. Mortality in these cases is often 100%, with no obvious environmental cause. In some species, there is considerable empirical evidence for a tightly regulated “death program”, or organismal initiation and control of the process leading to death.
Semelparity is common in long-lived plants, occurring in representatives of at least 20 different families [34]. In most cases, the proximate mechanisms of death are unknown, but there is evidence in some species for an internally generated “senescence signal”, which may be a hormone [35].
Semelparous animals show a similar pattern. Among vertebrates, the best studied examples of rapid senescence and death following semelparous reproduction are in fishes and marsupial mammals. Anadromous Pacific salmon (Oncorhynchus spp.), and catadromous eels (Anguilla spp.) and lampreys (Petromyzontidae) die shortly after spawning. Males of several small marsupials die shortly after mating (reviewed in [36]). Remarkably, the proximate molecular mechanisms of decline and death in all these distantly related vertebrates seem to be similar, with corticosteroid stress hormones playing the central role [37].
Among invertebrates, cases of death following semelparous reproduction that clearly seem to be coded into the genome (and thus “programmed”) include the many insects in which adults entirely lack mouthparts or are otherwise incapable of feeding [38]. In some cephalopod molluscs, rapid senescence and death follow spawning (squids) or brooding (octopuses). For example, in Octopus hummelincki, brooding females alter and reduce feeding behavior and die shortly after their eggs hatch. These traits are influenced by endocrine secretions from the optic gland [39,40].
In both plants and animals, life span can be is extended by experimental manipulations of physiology and/or reproduction, indicating that senescence is not the inevitable outcome of physical constraints. For example, lifespan of semelparous annual plants can often be artificially extended by removing reproductive structures [35]. Optic gland removal in O. hummelincki causes cessation of brooding, reinstates normal feeding behavior, and extends life [39]. In fishes, removal of the gonads (Pacific salmon, lampreys) or of the pituitary gland (lampreys) and prevention of mating (eels) can substantially increase life span [41-43]. Similarly in small marsupials, castration [44] or prevention of mating [45] can increase male life span up to that of females. These experiments indicate that rapid senescence and death following reproduction in at least some semelparous animals are genetically “programmed” and internally controlled.
Evidence from Iteroparous organisms
Iteroparous organisms have multiple reproductive episodes during the life cycle, as opposed to a solitary event. In these species, senescence leading to death is less abrupt and dramatic, but is ubiquitous nonetheless. Although data are difficult to collect, a high rate of death due to senescence in iteroparous species has been documented for some species even in the wild [46]. General explanations for organismal senescence are varied and contentious, and the idea that iteroparous senescence is “programmed” is not the majority view [5]. However, there is substantial evidence for genetic control of senescence in some iteroparous animals, indicating that senescence is at least sometimes “programmed” (in the inclusive sense) and highlighting the question of the relationship between aging “programs” and selection. For example, in Drosophila melanogaster, selective breeding and other studies demonstrated a genetic, heritable component to senescence [47-49]. In humans, geriatric diseases measured in some family members can predict life span in other family members [50]. In several model organisms, the specific genes influencing life span have been identified and there are now numerous examples where a molecular network promotes fitness early in life only to result in senescence and decreased fitness later (reviewed in [19]). These are the genetic mechanisms for the Anatagonistic Pleiotropy (AP) hypothesis discussed below. Many of the explanations can be attributed to the concept of disposable soma [51,52], which proposes a trade-off between investment in maintenance and repair versus investment in reproduction [53].
As one well-studied example, the IGF-1 signaling pathway is involved in regulating life span in several model organisms, including flies, nematodes, and rodents [54]. Several loci interact in what the authors term a “survival pathway”. The authors suggest that this pathway is activated to “survive unfavorable conditions in an effort to reproduce at a later time” (p. R664). The obvious implication is that the same survival pathway is deactivated under other conditions in which survival for later reproduction is less advantageous. Variants of the IGF-1 receptor homolog (daf 2) in the model organism C. elegans have produced empirical support for these arguments [55]. Recessive mutations in this gene significantly increased organismal lifespan. There are a number of other genes including those coding for steroids, apoptosis proteins and transcription factors that strongly support AP at the molecular level (for a review see [56]).
Evolutionary theory for organismal senescence
Most evolutionary explanations of organismal senescence are based on individual selection or selective neutrality, and interpret senescence as a byproduct rather than an adaptation selected for its own sake. Both the AP (discussed above) and the mutation accumulation hypotheses, two of the widely accepted explanations, propose mechanisms with negative effects that manifest late in life.
These hypotheses are distinguished by the early effects of genes and mutations. AP refers to traits with early positive effects that trade off against later negative effects [57,58] while mutation accumulation refers specifically to costly mutations with no fitness effect early in life [59,60]. Both of these hypotheses are based on the proposition that the force of natural selection declines with age, as the potential for further reproduction declines either abruptly or gradually [61]. The existence of genetic heritability in complex, functional pathways suggests that senescence is internally initiated and controlled by the organism. AP, therefore, meets our criteria for being “programmed” although in mutation accumulation a program per se seems unlikely. The failure of selection to maximize life span raises the question of why evolution would favor organisms that forego the potential advantage of longer reproductive lives.
The hypotheses described above are not mutually exclusive and are sometimes grouped as the “evolutionary” or “life history” theories of senescence [38,61]. There is a large body of literature, particularly for semelparous organisms, that is consistent with their predictions. In many semelparous organisms, the adaptive function of the nominal ‘death program’ is apparently to increase reproduction, with organismal death resulting as an incidental side-effect (making organismal senescence an example of “by-product death”, Table 1). In vertebrates, for example, the high levels of free corticosteroids that trigger rapid senescence and death also allow tissue protein to be consumed as an energy source during a brief and intensely competitive breeding season [62,63]. In Pacific salmon [64] and the marsupial Antechinus stuartii [36], the ability to exploit tissue protein as an energy reserve may increase reproductive success. In Octopus, the behavioral effects of optic gland secretion are likely to have an impact on hatching success, as time spent feeding is time not spent on brood care. In semelparous plants, a trade-off between survival and reproduction is supported by both experimental and comparative evidence. Various methods of preventing reproduction can extend life span [35], and reproductive effort is negatively correlated with life span across species [65].
The case for individual-based explanations of senescence is less clear for iteroparous organisms. Although some studies have supported life history theories of senescence, others have failed to provide clear support. For example, Williams (1957) predicted that “Low adult death rates should be associated with low rates of senescence, and high adult death rates with high rates of senescence” [57]. While this prediction has been empirically confirmed by some comparative studies of mammal populations [66,67], there are a few interesting exceptions that are worth mentioning. A comparison of guppy populations found the opposite pattern [68]. In interpreting the latter results as compatible with AP, Reznick et al. appealed to a different prediction generated from a mathematical model [69]. However, this interpretation has been criticized as misinterpreting Abrams’ work [70].
AP has also been tested in a long-running experiment using artificial selection for increased life span in Drosophila flies. After two years of selection, life span increased, while early fecundity declined, which corroborates the AP hypothesis [61]. After further selection however, life span continued to rise, while fecundity also rose at all ages, including early in life. While this result seems to contradict AP predictions it was instead interpreted by the authors as an experimental artifact [71]. Similarly, a recent study of Caenorhabditis elegans reported that selecting for early fecundity did not produce a cost in terms of longevity [72] while other studies designed to measure tradeoffs between survival and reproduction have found no evidence for them [73,74]. This criticism is not conclusive, however, and should be explored further as “incorrect predictions and faulty tests” have plagued studies of trade-offs in general [75].
Any cases for which we accept organismal senescence as being “programmed” (either in the weaker sense of being “coded” somehow in the genes or in the stronger sense of resulting from apparently goaldirected, hierarchical cascades of molecular-genetic interactions) challenge us to explain the origin and maintenance of the “death program”. AP and disposable soma hypotheses are widely accepted as the standard explanations for genetic programs that cause senescence in multicellular organisms [5]. However, there are some results (discussed above; see also [74]) that are in conflict with their predictions.
A common theme in all the hypotheses discussed above is that the organism is the primary unit of selection. In contrast, other authors have advanced alternative views for the adaptive value of POD in a multilevel selection framework. Among those authors that accept organismal death as programmed, many also support the hypothesis that death is adaptive, generally at a level of function other than the individual organism [17,19,30,70,74,76-78].
Supra-individual functions have occasionally been proposed for POD, although these suggestions sometimes lack both a rigorous theoretical framework and direct experimental evidence. The spectacular post-spawning mortality of Pacific salmon, for example, has been hypothesized to benefit offspring by enriching stream nutrient levels [19,79]. However, a benefit to offspring of the population at large, rather than of the dying individual specifically, would only be favored by a process of selection among populations, which has not been either documented or explicitly proposed. In other cases, benefits are apparently directed specifically to offspring of the dying individual, and this type of death could be favored by individual selection for maximizing successful reproduction. For example, in the neotropical rainforest tree Tachigalia versicolor, death of the parent tree is thought to open a light gap in the canopy that specifically benefits offspring of the dying tree [38,80].
It has been proposed that POD through organismal senescence is favored by kin selection [77,78], by group selection [19,70], or by selection among populations, with a more detailed model of the selection process than had accompanied early suggestions of population-level function [81].
Behavioral Suicide as Programmed Death in Animals
Decades ago, folklore held that arctic lemmings commit suicide in order to control population density for the good of the species. While this belief was widely held due to erroneous popular accounts, it was not empirically supported [82]. However, other observations of organismal suicide have been carefully researched and documented. Pea aphids (Acyrthosiphon pisum) that are parasitized by the Braconid wasp Aphidius ervi drop from the host plant, causing their own death. These insects live in groups of close kin, and their death prevents their body from producing parasites that would then attack their siblings. Death therefore increases the individual’s inclusive fitness, or equivalently, increases the fitness of the kin group [83,84]. At the same time, pleiotropic incidental mechanisms are also feasible. “Male self-sacrifice”, or apparent male cooperation with female cannibalism, is part of male copulatory behavior in several spider species, and in some insects [85,86]. One study also reported that males of the spider Argiope aurantia spontaneously die when they copulate [87]. Extensive details of the physiological or molecular mechanisms leading to “spontaneous” death or of the neurological mechanisms underpinning death-promoting behavior in spiders have not been reported. In redback spiders, cannibalized males increase their paternity compared to non-cannibalized males, so it is thought that self-sacrifice evolved due to this reproductive advantage [88]. The fitness benefit provided by this behavior raises intriguing possibilities. Among eusocial insects, workers routinely sacrifice their lives to benefit the colony and their siblings. One familiar example occurs in honeybee workers, which commit suicide each time they sting invaders in defense of their hive [89]. Similarly, these workers altruistically commit suicide in ways that apparently function to prevent the spread of disease within the hive [90]. In these cases, there is little mystery concerning function or evolution. Kin selection, or equivalently, selection among colonies, is a powerful force shaping the evolution of eusocial insects [91].
Programmed Death in Unicellular Organisms
Programmed cell death (PCD) serves the interests of multicellular organisms in many ways. Because it increases the fitness of the organism, there is no mystery about the function of PCD in plants and animals. It seems paradoxical, however, in cases where the single cell is the entire organism, so that PCD, by definition, is also POD. This issue is complicated by multiple competing criteria for what entities constitute organisms versus parts of organisms. In some species ambiguity arises as to whether cells are organisms or parts of organisms [2]. The apparent paradox of unicellular POD is common. POD with its diverse phenotypic manifestations has been observed in four of the five eukaryote super-groups defined by Keeling et al [92], as well as in prokaryotes [16,93] (also see Figure 1).
Evidence for POD in unicells
There is substantial evidence that cell death is often internally controlled and regulated in unicellular organisms, (Table 1 in [11]). The ancient and deeply conserved genetic and molecular pathways that cause cell death indicate that death-promoting mechanisms show less variability than would be expected for unavoidable or maladaptive death. For example, the caspase gene super-family is involved in cell death in both prokaryotes and eukaryotes, suggesting that it is functionally conserved and arose early in prokaryote evolution [20,94,95]. Potential convergence on the autophagic death phenotype also indicates that death-promoting mechanisms may have been selected for in unicells (though the function of the mechanisms may not be death per se in this example). The autophagic death phenotype and elements of its molecular pathway have been described in evolutionarily diverse, unrelated unicellular eukaryotes such as the stramenopiles [96], the kinetoplastids [97] and the social amoebae [98] (Figure 1). (Of course, one would need more evidence to rigorously establish that different mechanisms underpinning autophagy evolved independently and are maintained in different lineages by selective pressures stemming from similar environmental challenges.) Many bacteria undergo cell death mediated by hydrogen peroxide during stationary phase or in certain locations in a developing biofilm. Interestingly, gram-positive species produce hydrogen peroxide via the enzyme pyruvate oxidase [99] and references there, [16], while many gram-negative species generate hydrogen peroxide by the production of lysine oxidase [100]. The presence of (at least) two distinct means of triggering hydrogen peroxide production suggests convergent evolution, and thus suggests a widespread selective force favoring PCD in bacteria. The independent emergence of autophagic death in several eukaryotes and of hydrogen peroxide-mediated death in several prokaryotes is strongly supportive of POD as an adaptation although whether this is due to selection at levels other than the organism is debatable [11].
Examples of POD in model unicellular eukaryotes include yeast Saccharomyces cerevisiae [18,101-103], chlorophytes Chlamydomonas reinhardtii [29] and Dunaliella salina [33] and several protist parasites including Plasmodium [104], Trichomonas [105], Leishmania [106] and Trypanosoma [107,108]. Evidence of POD has also been documented in numerous non-model species including several eukaryotic phytoplanktons [15] (summarized in Table 1 in [11]). POD is also widespread among prokaryotes, where it occurs through a diversity of non-homologous pathways [16]. POD has been demonstrated in many bacterial species [100-110] including colonial cyanobacteria like Microcystis aeruginosa [111,112] and Microcystis flos-aquae [113].
Function and selection of POD in unicells
We must ask of unicellular organisms the same question we asked about multicellular organisms: Is death the function of POD, or merely a side effect? Multicellular organisms undergoing POD typically do leave offspring, so active death that is a byproduct of mechanisms “designed” to regulate investment in reproduction is a viable (and often supported) hypothesis for multicellular POD. However, unicells undergoing POD do not leave any direct descendants. Thus, in unicells, POD (with the possible exception of autophagy) may involve adaptation through function, and selection, at levels other than the individual organism. There is experimental evidence of benefits at levels other than the organism in bacteria and some eukaryotes, as well as a range of possible functions. Proposals include levels both below (genetic element) and above (group) that of the individual cell. Below we review evidence for benefits at levels other than the individual, as well as proposed functions and levels of selection for POD in unicells.
Evidence for benefits of POD in unicells
There is direct experimental evidence for higher-level benefits associated with POD in unicells. In bacterial species, several studies have compared group-level function in POD-positive versus PODnegative strains. Mai-Prochnow et al. showed that biofilms formed by a Pseudoalteromonas tunicata mutant defective in an autolysis pathway released far fewer propagules than the wild type [114]. The cells that were released from mutant biofilms were in a poor metabolic state, suggesting that autolysis may provide dispersing cells with nutrients as well as a means to escape the interior of the biofilm. The formation of healthy, physically stable biofilms in general is impaired in PODnegative mutants of Pseudomonas aeruginosa [115], Staphylococcus aureus [32,116], and Escherichia coli [117]. Furthermore, PODnegative mutants of Myxococcus xanthus are defective in fruiting body formation [118]. In E. coli, a small subpopulation of a clonal group undergoes autolysis, releasing anti-competitor toxins. In spatially structured environments, this strategy grants the clone as a whole enhanced competitive ability [119,120]. Another elegant study in E. coli allowed two strains, one capable of programmed death the other not, to compete against each other in the same environment [17]. On infection by a phage, the strain capable of programmed death outcompeted the non-POD strain, indicating that there is a fitness benefit at the clonal level despite the absolute fitness cost to the individual. This occurred even though individuals in the population were not closely related. These studies provide solid experimental evidence for grouplevel benefits of POD among bacteria. Evidence for benefits above the cell-level can be found in the eukaryote Dictyostelium mucoroides [121] and the chlorophytes Chlamydomonas reinhardtii [29] and Dunaliella salina [33]. The D. mucoroides life cycle resembles that of M. xanthus, and POD-negative strains are also unable to produce fruiting bodies. In C. reinhardtii, it has been shown that the mode of death impacts others in the population and that POD materials confer a fitness advantage on others while non-programmed death is harmful. Similarly, POD in D. salina allows others in the population to grow more vigorously and the mechanism was shown to be via dissolved organic materials (DOM) released during programmed death [33]. Interestingly, in this case a co-habiting archaeon could also utilize the liberated DOM. The significance of this in terms of the benefits being adaptive versus nonadaptive is uncertain.
Proposed Levels of Selection
Kin / group selection
As discussed above, there is considerable experimental evidence for fitness advantages from POD to clonal kin groups of cells of species usually considered to be ‘unicellular’. These fitness advantages may represent a significant source of selection pressure. As one well-studied example, clonal growth of yeast colonies creates opportunities for strong kin selection and evolution of group-beneficial altruistic traits [122]. PCD could serve at least two altruistic purposes in this context. By dying, a cell may spare nutrient resources for its neighboring kin group. Cell death could also release useful substances that can be used by the cell’s clone-mates [29,30,33,123]. In the model green alga Chlamydomonas reinhardtii, the benefit appears more active. In a direct POD vs non- POD fitness comparison, cells dying by POD enhance the fitness of relatives [29]. Furthermore, the fitness advantages of C. reinhardtii are species specific and cellular lysate following POD inhibits the growth of other Chlamydomonas species (manuscript in review).
Unicellular parasites are expected to be subject to strong selection among groups that share a host during a stage of the life cycle, and tend to have a shared fate as a result [124,125]. Consistent with this theoretical expectation, markers of POD have been described in many phylogenetically distant unicellular parasites. Furthermore, as in other unicells POD is active and requires protein synthesis. For example in Plasmodium, the inhibitory action of cycloheximide indicates that POD requires transcription and protein synthesis [104]. A direct group vs group comparison to assess group level effects of POD yielded unexpected results in Leishmania major [126]. Surprisingly, an inoculum containing apoptotic (cells dying by POD) organisms was more virulent than one without apoptotic organisms despite the inoculum containing apoptotic organisms having fewer viable parasites. The implication is that the apoptotic cells release substances that enhance the growth of others in the group.
Some groups of bacterial cells are functionally organized. In filamentous cyanobacteria such as Calothrix and Trichodesmium, poor conditions induce POD in some cells [127,128]. Because the filament often breaks up as a result, the function of POD in this system could be filament dispersal in response to unfavorable environmental conditions.
Biofilms are another common form of functional organization in groups of unicells. The lethal effect of chromosomal toxin-antitoxin systems may be mediated by extracellular signals [129], and these signals are central to biofilm development in Escherichia coli [117]. Homologous toxin-antitoxin systems have been identified in other gram-negative bacteria, gram-positive bacteria [130], and archaea [131]. A homolog to a phage holin-antiholin system on the chromosome of Staphylococcus aureus [132] suggests that these genes may serve a novel function at or above the level of individual cells. Homologs of this system are widespread among true bacteria and archaea [133].
In eukaryotes as well, the distinction between unicellular and multicellular is not always completely clear. For example, the social amoeba, Dictyostelium discoidium, is unicellular during part of its life cycle, but upon starvation, cells aggregate and develop into a multicellular motile slug, then fruiting body. Stalk cell death is programmed and is characterized by some classic features of POD [134]. The dead stalk cells function to support the spore head (which are often clonal relatives of stalk cells), presumably contributing to the spores’ dispersal. Therefore, the stalk cell death program appears to be the result of selection among multicellular aggregates. A different and less clearly functional death program has also been described in nonaggregated D. discodium [135].
Genic selection
The reproductive interests of genetic elements can differ from those of the organism carrying them. It is well established that gene-level selection can generate traits that are detrimental to individual fitness, while benefitting smaller genetic entities within them [136]. Indeed, genetic elements have been closely studied that routinely increase their inclusive fitness by killing their host [137]. In some cases, this peculiar form of selection may explain the function and evolution of genes for unicell POD. When a cell has low chances for reproduction, POD and the transfer of genes from the dying cell to a new host can save the genes. This process may be associated with occurrence of new mutations in the transferred genes and their improved ability to survive in the new host. This mechanism has been suggested to occur in S. cerevisiae [138] although the author allows that, “this hypothesis lacks direct experimental support”.
In prokaryotes, conjugative plasmids often encode for the simultaneous production of long-lived toxins and short-lived antitoxins [21]. When these are present together, the toxin is harmless to the host. However, if one daughter cell is cured of the plasmid, it will inherit toxic cytoplasm without a source of antitoxin and will die. While this is easily explained as a ‘selfish’ adaptation on the part of the plasmid, chromosomally-encoded toxin-antitoxin systems probably do not function in gene-level competition [139]. A similar puzzle arises in the case of holin-antiholin systems, which are encoded by bacteriophages and control the timing of host cell lysis and the dispersal of virions [140]. Chromosomally-encoded homologs are biochemically distinct from toxin-antitoxin systems, but also result in cell death (reviewed in [16]).
Plasmids and bacteriophages may induce cell death in the host (or non-host daughter) cell. Whether or not this is considered an example of POD, of course, hinges on whether or not selfish genetic elements are considered part of the organismal genome. The benefits in this case are straightforward and focused on the level of the selfish genetic element. This simple case may become more complicated when multiple levels of function and selection are involved. For example, in the case of phagemediated cell death in Pseudomonas aeruginosa, the lethal action of the lytic phage appears to confer a ‘group-level’ (or biofilm-level) benefit by allowing the release of a subpopulation of well-nourished disperser cells [114].
Population-level selection
The idea of selection among multi-generation populations or ‘demes’ predates more recent theory on selection among temporary ‘trait-groups’ [141]. Some authors have proposed that capacity for POD can offer long-term advantages to a species or other population. Because cellular damage is an important predictor of POD in yeast, it was suggested that one function of death is to remove cells with physical [142] or genetic [31,131,143] damage from the population. Elimination of mutated cells could help the population maintain its genetic stability. In “adaptive regrowth”, when the majority of cells in a stressed colony die by POD, a mutant, well-adapted minority arises and rescues the population, aided by the resources provided by lysed cells [30,101].
Conceptual Issues and Future Considerations
Given the definitions and usage of terms laid out at the start of this review, an answer to the title question as to whether internal, deathpromoting mechanisms can ever be adaptive, the answer must be a qualified ‘yes’. The evidence for programmed organismal death (POD) as a natural phenomenon is compelling for semelparous plants and animals, and substantial but more controversial for iteroparous macroorganisms. In these cases the ‘program’ (used in the inclusive sense here) appears to result largely from the constraints of physiological trade-offs as encompassed by, for example, antagonistic pleiotropy, notwithstanding some conflicting evidence. Explaining the evolution of death-promoting mechanisms often draws on lines of evidence that can be difficult or subtle to interpret (Table 2). However, if the death phenotype itself is adaptive, then selection at levels other than the individual organism is implicated. In cases where the death promoting phenotype is a by-product, then a complete explanation should ideally include both the adaptive function (purpose) of the mechanisms underpinning death as well as the reasons why selection has failed to achieve the function (purpose) without also promoting death. Death is a severe and absolute fitness cost. Except where there are physiological constraints as in antagonistic pleiotropy in multicellular organisms, the cases that seem to present the most pressing challenges are the welldocumented and diverse examples of POD in unicellular organisms. One intuitively appealing solution to this paradox is that the unit of adaptation for these traits in the unicellular world is a larger collective. Indeed the idea has arisen repeatedly that what we are accustomed to calling unicellular species are in some cases, and by some criteria, actually multicellular or at least colonial groups. Such a case can be made for eukaryotes such as social amoebae and other eukaryotes where group selection may play a significant role. In addition, for prokaryotes that form functionally important aggregates such as filaments and biofilms [130,144] the collective or group can sometimes be more relevant as a level of selection.
A valid criticism of this argument is that most of the evidence for POD as an adaptation in unicells rests, for the moment anyway, on the findings that POD can provide benefits to others. Nevertheless, group level advantages have sometimes been observed and POD in a unicellular colony may well provide inherited group level benefits due to population structures and genetic relationships that can be acted on by natural selection. However, as stated earlier benefits support but do not prove adaptation. For example, death in Pacific salmon may provide nutrients to developing embryos; but this form of POD can be explained by organism level fitness trade-offs.
The two classes of explanation that may have the widest applicability are those involving organismal adaptation under constraining tradeoffs (especially for macro-organisms), and those involving adaptation at higher levels of organization than the presumptive organism (especially for micro-organisms). In the latter cases, not only is a higher level of selection implied, but this selection must be strong enough to generate the extreme form of altruism represented by POD. Further progress on understanding the evolution of POD will depend on advancing diverse lines of evidence (Table 2) as well as considering carefully the effects of selection under constraining trade-offs and at multiple hierarchical levels. Such research is needed to fully reconcile the view that “every single organic being around us may be said to be striving to the utmost to increase in numbers” [1] with the observation that organisms can evolve mechanisms that promote their own death.
We gratefully acknowledge contributions by William W. Driscoll and Matthew Herron.

Tables and Figures at a glance

image   image
Table 1   Table 2


Figures at a glance

Figure 1
Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Article Usage

  • Total views: 13422
  • [From(publication date):
    July-2013 - Dec 08, 2019]
  • Breakdown by view type
  • HTML page views : 9546
  • PDF downloads : 3876