Jan Vijg and Paul Hasty
Mouse mutants displaying aging phenotypes much earlier in time than normal control animals offer the opportunity to develop and test interventions to reduce aging-related morbidity and mortality in humans. However, the few natural mouse mutants identified in the past as accelerated aging models have been criticized as less suitable for studying mechanisms ofnormal aging because neither the nature of the mutational defect nor the genetic background was accurately defined. A more general critique involved the perceived large variety of ways to reduce life span by causing pathologies similar to those normally occurring at late age. To address this problem one would ideally specifically design mouse models on the basis of known causes of aging. The validity of such models should then be demonstrated by showing that the genetic intervention accelerates a host of symptoms of normal aging rather than one or few.
Accumulation of DNA damage with age has been implicated as a major cause of aging, an hypothesis strongly supported by the discovery that most human segmental progeroid syndromes are caused by heritable mutations in genes involved in DNA repair and genome maintenance. There are now a number of mouse models, harboring engineered defects in various aspects of genome maintenance, which prematurely display a range of aging phenotypes. This strengthens the hypothesis that aging is driven by spontaneous DNA damage causing genomic instability and cellular stress responses, such as apoptosis and cellular senescence, eventually resulting in systems dysfunction at all levels. In this chapter we will review the recent developments in this field and discuss the validity of the aging pheno-types observed in the mouse models, with a focus on the possible implications with respect to DNA damage as a proximate cause of aging common to all mammals.
Aging is poorly defined at the mechanistic level, even though in recent years much progress has been made by studying genetic mutations altering life span and the onset of age-related characteristics (Vijg and Suh, 2005). Subjects of these analyses include yeast (Saccharomyces cerevisiae), nematodes (Caenorhabditis elegans), fruit flies (Drosophila melanogaster), and mice (Mus musculus). Intriguingly, in all these species, mutations downregulat-ing activities of growth, reproduction, or nutrient sensing significantly increase longevity, possibly by interfering in the generation of somatic damage or through the upregulation of mechanisms that protect against such damage.
Although the mechanisms underlying age-related degeneration and death in these and other species still need to be uncovered, a consensus as to why and how we age has begun to emerge. It is now generally accepted that the time-dependent decrease in fitness in most multicellular organisms is nonadaptive; that is, it is not controlled by a purposeful genetic program similar to the control of development. Aging provides no specific advantage to the individual and most researchers now accept that age-related degeneration and death is ultimately due to the greater relative weight placed by natural selection on early survival or reproduction than on maintaining vigor at later ages. This decline in the force of natural selection is largely due to the scarcity of older individuals in natural populations owing to mortality caused by extrinsic hazards (Kirkwood, 2005). Because resources are limited, this strategy would not allow a maximization of somatic maintenance and repair, which is not required for periods of time that greatly exceed the time needed to reproduce. Under conditions of lower extrinsic mortality, permitting reproduction at later ages, the allocation of resources would shift toward somatic maintenance, thereby increasing life span. However, maintenance of the soma is never maximized since, according to the nineteenth-century biologist August Weismann, the soma merely provides the housing for the germline, seeing to it that the germ cells are protected, nourished, and conveyed to the germ cells of the opposite sex to create the next generation (Kirkwood and Cremer, 1982). With the soma being dispensable, the trade-off between growth and reproduction on the one hand and somatic maintenance on the other, is biased toward reproduction.
This ''disposable soma theory'' not only provides a rationale for why we age, but also predicts the nature of its proximate cause; that is, the accumulation of unrepaired somatic damage (Kirkwood, 2005). This explanation, which is now supported by a large body of evidence, also explains the similarities in symptoms of aging, both within and across species, and the apparent universality of genetic pathways of life extension across different phyla. Indeed, rather than being programmed to age, animals are programmed to survive long enough to reproduce, possibly by using highly conserved cellular defense systems against somatic damage common to all or most species. Such damage may come from the environment e.g., radiation, infectious agents but also from inside the organism e.g., reactive oxygen species (ROS), normal by-products of metabolism.
Although it is now clear that life span is highly plastic and can be manipulated by metabolic switches that can affect the levels of spontaneous somatic damage or the proportional effort that is devoted to somatic maintenance, we still have only limited insight into the mechanisms of aging in different animal species. To elucidate human aging, mice are a good model system for several reasons. First, mice are positioned close to humans on the evolutionary scale. Second, their relatively short life span and small size permit extensive life span studies on an economic basis. Third, mouse genetics has closely emulated the progress in human genetics and is now almost equally powerful. Fourth, although in mice a full phenotypic characterization of aging is still far from the systematic catalogue of signs and symptoms of old age presently available for humans, the species ranks a solid second with rapid improvements underway.
Indeed, in the wake of the current explosion in genetically engineered mouse models, major coordinated efforts have emerged to obtain standardized and comprehensive databases for morphologic, biochemical, physiologic, or behavioral characteristics of various mouse strains (e.g., the mouse phenome project; http://www. jax.org/phenome). More recently, patterns of age-related pathology of the mouse have been made computationally accessible using ontologies—controlled vocabularies of terms—in the context of a federated database (e.g., www.niehs.nih.gov/cmgcc/dbmouse.htm). Such progress greatly facilitates the development and use of mouse models of accelerated aging.
Mouse models of accelerated aging have been criticized in the past based on the argument that many of the degenerative phenotypes associated with aging could result from a variety of interventions and need not necessarily involve the same causes that underlie natural aging. Although this is a valid argument, it should be realized that the use of model systems to study natural phenomena is a generally accepted approach in biology. For example, human cancer has been studied extensively using laboratory rodents subjected to treatment with a variety of genotoxic agents. Although we were all well aware of the fact that natural human cancers normally were not caused by such treatments, this approach nevertheless allowed us to obtain valuable information about the etiology of this disease. Of note, such rodent models for studying human cancer were based on the rationale that cancer is caused by DNA damage—hence, the use of DNA damaging agents to generate these model systems. By the same token, the most recent series of mouse models for human cancer is based on highly specific genetic alterations, known to increase human susceptibility to cancer. Hence, a logical approach in generating animal models for human aging is to develop specific interventions based on our increased knowledge of what causes human aging.
What can we say about the proximate causes of aging? The disposable soma theory predicts that aging is caused by the accumulation of unrepaired somatic damage (Kirkwood, 2005). There are strong arguments that DNA damage is the most important type of age-accumulated damage and a likely cause of many aging-related phenotypes. Among biological macromolecules, the DNA of the genome is unique in view of its role in transferring genetic information from cell to cell and from generation to generation. A strong, logical argument to consider the DNA of the genome as the Achilles heel of an aging organism is the lack of a back-up template. This is in contrast to proteins, which at least in principle, can be easily replaced with the corresponding gene as template. Indeed, the maintenance of genomic DNA is of crucial importance to survival because its alteration by mutation is essentially irreversible and has the potential to affect all downstream processes. A logical approach for making mouse models for human aging, therefore, would be to inactivate genes involved in DNA repair and genome maintenance. Interestingly, nature itself has preceded us, in this respect, by creating natural human mutants displaying premature aging as a consequence of defects in DNA repair and genome maintenance. This is by itself a powerful argument that changes in DNA drive the aging process. Indeed, genetic defects in few, if any, other systems than DNA repair and genome maintenance have been associated with premature aging (Martin, 2005). The rationale for generating mouse models of human aging on the basis of genetic alterations in genome maintenance pathways is therefore strong but not without problems.
Next we will first discuss the validity of premature aging symptoms in such mouse models and then discuss a number of them in more detail.
Aging differs from all human diseases by its complexity. It is the most complex phenotype currently known and the only example of generalized biological dysfunction. Its effects become manifest in all organs and tissues, it influences an organism's entire physiology, impacts function at all levels and increases susceptibility to all major chronic diseases. Nevertheless, typical symptoms of aging, often similar across species, can and have been defined.
For human aging, valuable information has been gleaned from a century of clinical observations. It was on this basis that, as mentioned earlier, a series of life-shortening genetic alterations in humans were described over a century ago that appeared to accelerate multiple signs of normal aging (Martin, 2005). These so-called segmental progeroid syndromes, already briefly discussed, were described by the medical community well before the discovery of DNA, and are therefore not biased toward a DNA-based hypothesis of aging. So it is remarkable that so many of these syndromes are defective in genome maintenance. The most striking of the human progeroid syndromes are Werner Syndrome (WS) (Epstein et al., 1965) and Hutchinson-Gilford Progeria Syndrome (HGPS) (Pollex and Hegele, 2004). WS is caused by a defect in a gene that is a member of the RecQ helicase family (Yu et al., 1996). The affected gene, WRN, encodes a RecQ homologue whose precise biological function remains elusive, but is important for DNA transactions, probably including recombination, replication, and repair.
HGPS is caused by a defect in the gene LMNA, which through alternative splicing encodes both nuclear lamins A and C (Eriksson et al., 2003). Nuclear lamins play a role in maintaining chromatin organization. Less striking segmental progeroid syndromes include ataxia telangiectasia, caused by a heritable mutation of the gene ATM (ataxia telangiectasia mutated), a relay system conveying DNA damage signals to effectors (Shiloh, 2003), Cockayne syndrome and trichothiodystrophy, diseases based on defects in DNA repair and transcription (Lehmann, 2003), and Rothmund Thomson syndrome, like Werner syndrome, based on a heritable mutation in a RecQ gene (Lindor et al., 2000). There is evidence that each of these genes when defective can also lead to aging symptoms in the mouse, sometimes in combination with other gene defects (see later, and Hasty et al., 2003).
Although in both humans and mice cancer incidence increases exponentially with age, the tumor spectrum in the two species differs significantly, with sarcomas and lymphomas predominant in the mouse and epithelial cancers in older humans (DePinho, 2000). Likewise, the spectrum of normal age changes (other than cancer) in mice and humans is not exactly the same, which always needs to be kept in mind when using these models (Hasty and Vijg, 2004). Moreover, although cancer as a phenotype is generally undisputed, aging has diffuse characteristics, and includes cancer and a variety of degenerative phenotypes (see www.niehs.nih.gov/ cmgcc/dbmouse.htm). Progeroid genotypes are associated with an early onset of some, but not all, characteristics of senescence and must therefore be interpreted with caution.
Loose criteria that help identify genuine mouse mutants of accelerated aging are (1) the phenotype should present after development and maturation are complete; (2) the phenotype should be demonstrable in control populations at a more or less similar point in their survival curve; and (3) the genetic alteration should accelerate multiple aging phenotypes (Hasty and Vijg, 2004). None of these criteria is written in stone. Indeed, accelerated aging can occur even before development is complete, as in the case of HGPS. Such cases, however, are more difficult to recognize as authentic models of aging and may not be as valid as those that exhibit aging phenotypes after maturation. It is also easily imaginable that a genetic alteration accelerates certain symptoms of aging much more than expected on the basis of the survival curve. Such so-called exaggerated aging would be expected if, rather than a quantitative, chronological master switch, the mutation would affect only one critical pathway for somatic maintenance leading to severe imbalance of the survival network.
Interestingly, the single-gene mutations that increase life span in worms, flies and mice may do so through the upregulation of cellular defense systems, including DNA repair and antioxidant defense. Candidate genes implicated in the control of such a survival response are FOXO and SIRT1, which have been demonstrated in nematodes and fruit flies to control downstream targets of the pro-longevity mutations affecting nutrient sensing, reproduction, and growth (Vijg and Suh, 2005). Down-regulation of these effector genes could then conceivably lead to an acceleration of all possible aging phenotypes. FOXO3a and SIRT1 knockout mice do not display apparent signs of accelerated aging, although it is possible that a progeroid phenotype will become visible after a more quantitative downregulation of these genes (Cheng et al., 2003; Hosaka et al., 2004).
It should be noted that the mutations that lead to increased longevity in nematodes, flies or mice are likely to do so only at the cost of some selective disadvantage, often not obvious under laboratory conditions (Jenkins et al., 2004). For some of the mouse longevity mutants, such as the growth hormone deficient Ames dwarf mice, fitness costs are readily apparent in the form of infertility and hypothyroidism (Bartke and Brown-Borg, 2004). However, for another longevity mutant in the mouse, p66SHC, there is no obvious selective disadvantage
(Migliaccio et al, 1999). At this time the only known consequence of deleting p66SHC is increased life span; thus, we presume the laboratory environment masks any disadvantage. For example, it is possible that p66SHC functions to increase cellular ROS to initiate cellular destruction as a part of our defense system against infectious agents. This potential disadvantage would be masked in the p66SHC-mutant mice since they are housed in a pathogen-free environment. We currently lack the detailed phenotypic comparisons to confirm that longevity-conferring mutations do so by retarding all possible symptoms of aging equally. Hence, though it is clear that mutations in single genes can activate survival pathways conferring increased longevity, the concept of master regulator genes to control the rate of aging is doubtful.
Defects in Genome Maintenance: Observations for Aging and Cancer
Embryonic stem cells and gene targeting technologies have been instrumental in confirming during the last decade the general prediction that genome maintenance systems are critical for suppressing tumor formation (Hoeijmakers, 2001). The data show that mutation of a single gene can increase genomic instability, leading to cancer. Of note, the cancer spectrum in such mutants, as in human cancer hereditary syndromes, is segmental even though these DNA repair pathways function in many cell types that are not predisposed to cancer. Importantly, while focused on cancer, several investigators have found that mutating some genome maintenance genes causes progeroid syndromes. Currently there are a number of mouse lines harboring specific genetic alterations that present with a shortened life span and precocious aging phenotypes (for an exhaustive listing, see Lombard et al. (2005)). In each of these mouse models some aspect of genome maintenance is affected. Next, we will briefly discuss the most prominent models with some details of the aging-related phenotypes they affect and their possible link to one or more causal factors. It should be noted that none of these models has been subjected to standardized, objective phenotyping, only some of them were compared side-by-side with littermate controls, and the genetic background may be different from model to model, making comparisons difficult.
Defects in cell cycle control have been demonstrated to cause premature aging in mice, most notably mutations in the p53 gene. The p53 tumor suppressor is a transcription factor that controls a network of genes that regulates responses to DNA damage (Vogelstein et al., 2000). The p53 protein inhibits cell cycle progression, facilitates DNA repair, and activates both cellular senescence and apoptosis pathways (Campisi, 2005). Two mutant alleles of p53 have been described, which, in combination with a wild-type allele, result in multiple symptoms of aging and a shortened life span. The p44 allele, a naturally occurring, splice varient first described in 1987, lacks the first transactivation domain (Maier et al., 2004). The other mutant allele, termed M, is a deletion of the 5' region of the p53 gene, lacking both transactivation domains (Tyner et al., 2002). Both mutants display prominent signs of premature aging, but the two animal models differ, mainly in the severity of the symptoms. The first symptoms of aging in the M mutant become apparent only at 18 months, whereas the p44 animals already show aging-related mortality as early as five months. Severity is probably correlated with the expression level of the mutant allele, which is very low in the case of the M allele. In the p44 model it has been demonstrated that with increasing p44 expression symptoms become much more severe.
Both models display early osteoporosis and kyphosis, major forms of aging-related pathology in both mice and humans. The p44 mice show early fertility loss, to some extent caused by a breakdown of the reproductive axis (H. Scrable, personal communication). This loss of the hypothalamic pituitary gonadal axis is very similar to the situation in natural human and mouse aging (Wise et al., 1997). This loss of fertility is not observed in the M mutant, but again, the symptoms in that model are generally much less severe. Both mouse models suffer from a typical aging-related redistribution of fat resulting in loss of subcutaneous fat. Other typical aging-related phenotypes in these mice are skin problems and various forms of atrophy.
Surprisingly, in both mutants cancer incidence was lower than in the controls. Of note, mice expressing additional copies of wild-type p53 under the control of its own promoter do not show signs of premature aging, but do show lower cancer incidence (Garcia-Cao et al., 2002). In this case, the normal p53 gene dosage is minimally increased (i.e., by one additional copy), resulting in an increased response to DNA damage without affecting basal levels of p53. By contrast, in the two accelerated aging mutants, p53 may be constitutively activated through an interaction of a truncated p53 with a full-length polypeptide. This apparently results in postnatal growth impairment of the animals, possibly caused by a reduced cell proliferation rate as observed in cultured fibroblasts from these animals (Maier et al., 2004). It is also supported by the observation of a two-fold increase of senescent cells in tissue sections of liver and spleen in the p53+/M mice, as compared to control animals (Dumble et al., 2004), a modest rise in comparison to the increase in the number of senescent cells in both mutant and control animals during aging, which was more than ten-fold in these tissues. (Cellular senescence is the irreversible cessation of cell division, which can be observed with all normal mammalian cells in culture.
Senescent cells can be identified by staining for ^-galactosidase at pH6.0.)
Of note, increased rates of apoptosis (programmed cell death) in these two mutant mouse models have not been observed (H. Scrable and L. Donehower, personal communication). Indeed, the hyperactive p53 protein most likely causes its effects through increased cell cycle arrest, leading to a general inhibition of cell proliferation. This may explain many of the premature aging pheno-types. For example, impairment of osteoblast proliferation, a likely natural cause of osteoporosis in the elderly, may cause the increased, premature osteoporosis in these mutant mice.
Overall, therefore, abnormally enhanced p53 activity appears to promote many aging-related phenotypes through the inhibition of normal regenerative processes that are essential for adult animals to survive and maintain organ function. As the guardian of the genome, p53 is supposed to inhibit cell growth and proliferation, but only in response to DNA damage, when extra time is needed for repair. Though cancer is also a major aging-related phenotype, enhanced p53 activity would be expected to greatly suppress tumor formation due to its inhibition of normal cell proliferative activities. It is conceivable that in normal mice p53 responses naturally increase with age due to the aforementioned increased load of DNA damage. This would then result in very similar phenotypes as in the p53 mutants, but later in life, as part of the normal aging process.
Was this article helpful?