Special Reviews

Aging and Genome Maintenance: Lessons from the Mouse?

See allHide authors and affiliations

Science  28 Feb 2003:
Vol. 299, Issue 5611, pp. 1355-1359
DOI: 10.1126/science.1079161


Recent progress in the science of aging is driven largely by the use of model systems, ranging from yeast and nematodes to mice. These models have revealed conservation in genetic pathways that balance energy production and its damaging by-products with pathways that preserve somatic maintenance. Maintaining genome integrity has emerged as a major factor in longevity and cell viability. Here we discuss the use of mouse models with defects in genome maintenance for understanding the molecular basis of aging in humans.

The accumulation of somatic damage is now considered a main cause of the aging process in species varying from nematodes and insects to mice and humans. Among the various sources of somatic damage, reactive oxygen species (ROS), the natural by-products of oxidative energy metabolism, are often considered as the ultimate cause of aging (1). However, free radicals also participate in physiological processes that benefit fitness, such as growth factor signal transduction (2). Thus, optimal energy production must be balanced against the damaging effects of ROS. This trade-off is highlighted by several recent reports on mouse mutants with extended life-span. Homozygous inactivation of the p66 isoform of the Shc gene (3), which controls stress-induced ROS and apoptosis, leads to significant life-span extension together with increased resistance to ROS (4). Similarly, heterozygous inactivation of the gene for insulin-like growth factor I, important for growth and metabolism, significantly lengthens the life-span of mice and increases resistance to ROS (5). Interestingly, this possible trade-off between growth and reproduction and somatic maintenance is in keeping with the suggestion that caloric restriction may extend life-span by attenuating the oxidative stress caused by normal metabolism (6). In general, there is a price to be paid for such unnaturally high somatic maintenance, and that price is reduced reproductive success. This is not necessarily apparent in laboratory settings but can be revealed by simulating more “natural” environments, as recently seen with life-span extension mutants of the nematode Caenorhabditis elegans (7). What is the route from ROS to the adverse effects associated with aging? Among myriad possibilities, alterations in the genome have long been considered critically important. Recent results obtained with mouse models that show accelerated rather than retarded aging as a consequence of defects in genome maintenance systems support this view and are beginning to shed light on the question of how we age.

One of the cellular targets of ROS is DNA. More than 100 different types of oxidative DNA lesions have been described, ranging from base modifications to single- and double-strand DNA breaks and interstrand cross-links (8). These lesions disrupt vital processes such as transcription and replication, which may cause cell death or growth arrest or may induce mutations that lead to cancer. To cope with DNA damage, organisms evolved an intricate network of DNA damage repair pathways, each focusing on a different class of lesion (8). Excision systems deal with lesions that affect only one DNA strand, which permits excision of the lesion and subsequent use of the intact complementary strand to fill the gap. Base excision repair (BER) and nucleotide excision repair (NER) remove subtle and more helix-distorting types of damage, respectively. They operate genome-wide and are important for preventing mutations. Transcription- coupled repair (TCR), on the other hand, eliminates lesions that actually block the transcription machinery, thus helping to repair those genes that are actually used by the cell. TCR uses many components of NER. BER, TCR, and (to a lesser extent) NER are critically important for repairing ROS-induced base damage and single-strand breaks. ROS can also induce DNA double-strand breaks, which are repaired by homologous recombination when another intact DNA copy is available (after replication) or by nonhomologous end joining. Finally, interstrand cross-links, another double-strand lesion that can be induced by ROS, are eliminated in a poorly understood fashion probably involving homologous recombination and some NER components. DNA lesions also cause problems during replication, a time of great vulnerability for the cell. Some of the damage incurred during replication is bypassed by a special class of translesion polymerases at the expense of a higher risk of mutations. Recombination systems presumably involving RecQ-like helicases as well as other proteins also play a role in coping with DNA lesions during replication. Do any of these genome maintenance systems prevent age-related phenotypes by ameliorating or repairing DNA damage caused by oxidative metabolism? Natural human mutants provided the first evidence that they do, but these mutants had been discovered long before the discovery of DNA repair.

Human Segmental Progeroid Syndromes

In humans, several heritable mutations accelerate the onset of multiple aging phenotypes (Table 1) (9). Disorders caused by these mutations are termed segmental progeroid syndromes, because they accelerate some but not all signs of normal aging. The most prominent of the currently known premature aging syndromes were described a century ago, about the time DNA was first described by Miesher in his first attempts to identify the chemical makeup of cell nuclei. By now it is known that many of the heritable mutations giving rise to the symptoms of accelerated aging in these patients partially or wholly inactivate proteins that sense or repair DNA damage. This finding suggests that failure to maintain genomic integrity underlies at least some aging phenotypes. Although not all heritable defects in DNA repair display symptoms of accelerated aging, it is striking that so many progeroid disorders are linked to defects in genome maintenance.

Table 1

Most commonly described human segmental progeroid syndromes.

View this table:

The mouse could prove to be a powerful model for understanding the importance of genome maintenance in human longevity. The genomes of mice and humans are similar and genome maintenance mechanisms are generally highly conserved. Mouse mutants defective in genome maintenance are now available and, like their human counterparts, many display segmental progeria (Table 2). Below, we review the current repertoire of mouse models for genome maintenance defects that also display symptoms of accelerated aging. We have organized our discussion according to the major genome maintenance systems for removing DNA damage induced by ROS and, when possible, have compared mice and humans.

Table 2

Aging-related phenotypes in mouse models with genome maintenance defects. KO, knockout; HM, hypomorph; m, dominant active mutation; nr, not reported but may not have been investigated; no, not observed after investigation; topo in, topoisomerase inhibitors; IR, ionizing radiation; MMC, mitomycin C.

View this table:

DNA Excision Repair

Among the mouse models with mutations in DNA excision repair, only those that affect both NER and TCR show symptoms of accelerated aging. Germline deletion in mice of any gene that is essential for BER is embryonic lethal (10), which suggests that this type of DNA damage is very frequent and must be removed swiftly. On the other hand, germline deletion of one of the many glycosylase genes, which repair only a subset of BER substrates, generally results in no obvious phenotype (11), probably as a result of functional redundancy. Hereditary defects in essential BER components have not been found in humans, most likely because they are embryonic lethal.

In contrast, NER deficiencies are associated with three human heritable disorders: xeroderma pigmentosum (XP), trichothiodystrophy (TTD), and Cockayne syndrome (CS). Individuals with XP can be classified into at least seven complementation groups (denoted XPA through XPG). Relative to unaffected individuals, they exhibit a >1000-fold increase in ultraviolet B (UVB)–induced skin cancer, accelerated photo-aging of the skin, and frequently neurodegeneration. Mice deleted for XPA are devoid of NER and, similar to human patients, show increased susceptibility to skin cancer (12). These NER-deficient mice develop normally, arguing against a major role for NER in repairing spontaneous oxidative lesions.

The other two NER-related disorders, TTD and CS, display prominent symptoms of accelerated aging. TTD individuals suffer from neurological and skeletal degeneration, cachexia, ichthyosis, and characteristic brittle hair and nails. TTD is caused by point mutations in the XPD gene, different mutations in which can give rise to XP, CS, or TTD. XPD encodes one of the two core transcription factor IIH (TFIIH) helicases. TFIIH facilitates partial unwinding of the DNA duplex and is required for transcription initiation by the RNA polymerase I and II transcription machineries and for repairing DNA lesions by NER (7). To model the XPD mutation that causes TTD, de Boer et al. (13) generated an analogous mutation in mice (14). Cells from these mice are transcriptionally impaired and exhibit only 20 to 40% DNA repair synthesis after UV irradiation. The mouse model closely resembles the human disorder. TTD mice prematurely show age-related phenotypes, including osteoporosis, osteosclerosis, gray hair, cachexia, and reduced life-span (Table 2) (13). However, in contrast to the human equivalent, TTD mice also show increased susceptibility to UV or chemically induced cancer, although they are not as cancer-prone as XPA-deficient mice, which are totally deficient in NER.

In view of the multifunctionality of XPD, the premature aging of TTD mice could be due to defective NER/TCR or impaired transcription. It is unlikely that accelerated aging is solely due to the NER defect, because XPA patients and xpa –/– mice, which are completely NER-deficient, are highly cancer-prone but do not exhibit pronounced aging (12). However, when the partial repair defect in TTD mice is converted to a total NER defect by crossing them to XPA mutant mice, the aging phenotype is greatly exacerbated (13). This suggests that a complete NER defect, in combination with a TCR defect that renders the transcription machinery very sensitive to lesions in transcribed genes, is responsible for the markedly accelerated aging of the double mutant mice.

Only both mutations together lead to a measurable increase in the sensitivity of cultured cells to damage by paraquat or ionizing radiation. This observation suggests that a subset of ROS-induced lesions initiates the age-related decline in TTD mice, because paraquat, ionizing radiation, and ROS generate overlapping DNA lesions. Thus, XPD deficiency likely impairs TFIIH function, which compromises both repair and transcription. In addition, increased genomic instability, suggested by the increased cancer predisposition, may also contribute to the aging phenotypes.

Whereas human TTD is caused by partial defects in NER/TCR and transcription, CS is caused by a defect in only TCR. CS is caused by mutations in either CSA or CSB, proteins whose exact biochemical functions are uncertain. CS individuals exhibit selected aging-related phenotypes, such as cachexia, neuronal degeneration, and loss of retinal cells. Like TTD, CS does not increase susceptibility to cancer. CS mouse models are not as severely affected as humans, but as they age, they develop neurological abnormalities, including tremors, limb ataxia, and inner ear defects. They also show cachexia and retinal degeneration (15). Like the mouse model mimicking TTD, CSB mutant mice show a modest increase in cancer susceptibility and a pronounced exacerbation of the CS aging symptoms when crossed to XPA mutant mice (16).

Repair of Double-Strand Lesions

One of the most severe types of DNA lesions that can be induced by ROS is a double-strand lesion, such as a double-strand break or interstrand cross-link. Such lesions very frequently lead to genome rearrangements, including translocations, inversions, and large deletions. In mammals, cancer is frequently associated with large genome rearrangements, and there is general consensus that such mutations frequently arise from inaccurate repair of double-strand lesions. Double-strand lesions are also very cytotoxic, hence the success of radio- and chemotherapeutic agents that induce such lesions in cancer cells. The pathways that repair double-strand breaks are homologous recombination and nonhomologous end joining (NHEJ). Segmental human progeroid syndromes, such as ataxia telangiectasia and Werner syndrome, are likely due to defects in the signaling or cellular responses to double-strand lesions. However, to date, no human disorder resulting from an absolute defect in a double-strand break repair pathway has been identified. Several mouse models with defects in genes for homologous recombination repair have been made, but thus far none have shown obvious aging-related symptoms (17). However, some of these mutations result in embryonic lethality, and very few mice with mutations in genome maintenance that survive have been carefully studied over their entire life-span, so a definite conclusion regarding potential aging phenotypes in such mice is premature at this stage.

By contrast to recombination, mice with a defect in a NHEJ gene, Ku80 (Ku86), exhibit an early onset of aging characteristics. NHEJ involves Ku80 and Ku70 (which form a heterodimer called Ku), the ligase IV-XRCC4 complex, DNA-PKcs, and Artemis (8, 18). Mice deleted for any of these genes exhibit classical signs of impaired NHEJ, including defective V(D)J recombination, hypersensitivity to ionizing radiation, and increased genomic instability (19, 20). Failed V(D)J recombination causes severe combined immunodeficiency, so thatku80 –/– mice can only survive in a pathogen-free environment, which explains why NHEJ-deficient humans are rare and need bone marrow transplantation to survive (18). The Ku80 mutant mice exhibit several characteristics of aging, including osteoporosis, growth plate closure, atrophic skin, liver pathology, sepsis, cancer, and shortened life-span (Table 2) (21). Deletion of Ku80 also increases sensitivity to ROS (22).

DNA breaks, genomic instability, and apoptosis are directly related to oxidative metabolism in ku80 –/–mice and cells (23). Thus, ROS-induced genomic instability may drive the premature aging phenotypes. Although Ku80 is essential for NHEJ, it is uncertain whether defective NHEJ is the primary cause for premature aging. Defects in another NHEJ protein, DNA-PKcs, do not cause premature aging in mice. In yeast, Ku80 is additionally important for maintaining telomeres and repressing subtelomeric chromatin (24). Mammalian Ku80 may also be important for telomere maintenance because it localizes to telomeres (25) and prevents telomeric fusions (26). Thus, Ku80 may postpone aging phenotypes by promoting NHEJ and/or maintaining telomeres. Mice with inactivated Ku70 do not seem to suffer from accelerated aging, but they do show increased spontaneous tumor formation (27).

A second example of accelerated aging as a consequence of a defect in the repair of double-strand lesions involves ERCC1, which acts as an endonuclease in both NER and interstrand cross-link repair (8). Most ercc1 –/– mice die about 3 weeks after birth from liver and kidney abnormalities that are reminiscent of accelerated aging, including premature polyploidization and appearance of nuclear inclusions (28). These animals also exhibit other age-related phenotypes that include cachexia, neuronal degeneration, and skin atrophy (Table 2). These phenotypes generally develop in the second week of life and progress rapidly. A human equivalent for the ERCC1 mouse mutant was recently discovered in a patient with a mutation in the XPF gene, the product of which functions in a complex with ERCC1 (8). This patient also suffered from a prematurely aged appearance, exhibiting progressive liver and kidney dysfunction, cachexia, hypertension, neuronal degeneration, and skin atrophy (29).

Telomere Maintenance

Telomeres are the repetitive sequence and specialized proteins that cap the ends of linear eukaryotic chromosomes, protecting them from degradation, fusion, and recombination (30). Telomeres have been proposed to contribute indirectly to mammalian aging because critically short or dysfunctional telomeres induce cellular senescence (cells maintained in a dysfunctional, permanently arrested state and no longer able to proliferate). Telomeres may also participate in DNA repair because they bind proteins such as Ku80 (25). The importance of telomere function in age-related pathology is demonstrated in mice lacking the telomerase RNA component (mTR) (31). Because laboratory mice have very long telomeres, with lengths three to five times those of humans, early-generation mtr –/– mice have no obvious phenotype. However, with successive generations, the telomeres ofmtr –/– mice shorten. After three or four generations, organs with proliferating cells fail (32) and selective signs of premature aging develop (33). These signs include gray hair, alopecia, skin ulcerations, impaired wound healing, cancer, and shortened life-span (Table 2). In addition, chromosomal fusions, a sign of genomic instability, develop more rapidly in mtr –/– mice, requiring 15 as opposed to 24 months (in wild-type mice) to accumulate to appreciable levels (33). Interestingly, deletion of ATM (the gene mutated in human patients with ataxia telangiectasia, a segmental progeroid syndrome) exacerbates the aging phenotype of telomere dysfunctional mice (34). An aging phenotype had not been previously reported inatm –/– mice, which exhibit a high cancer incidence and mild neurodegeneration. Dysfunctional telomeres greatly reduced cancer for theatm –/– mice, perhaps allowing the aging phenotype to become manifest. More likely, excessive long telomeres that are common for mice but not for humans have alleviated the aging phenotype in atm –/– mice.

RecQ Helicase Suppressors of Genome Instability

RecQ-like DNA helicase genes are best known for their mutant forms that cause Bloom, Werner, and Rothmund-Thomson syndromes in humans. Of these three disorders, Werner syndrome (WS) is generally considered the best example of premature human aging because of its adult onset and striking resemblance to normal aging (9). WS individuals prematurely develop atrophic skin, thin gray hair, osteoporosis, type II diabetes, cataracts, arteriosclerosis, and cancer, and WS cells exhibit genomic instability (Table 1). Typically, WS individuals die in the fifth decade of life, primarily of cardiovascular disease or cancer. However,wrn –/– mice have no obvious aging phenotype (Table 2) (35), possibly because the role and regulation of the WRN gene (36) differs between mouse and human. WRN is a homolog of the Escherichia coliRecQ helicase (37) that suppresses undesirable recombination (38) and copurifies with a DNA replication complex (39) and components of NHEJ including Ku80 (40). Therefore, the aging phenotype observed forku80 –/– mice may be mechanistically similar to WS in humans. In addition, wrn –/– mouse cells exhibit a phenotype that resembles WS human cells, including increased spontaneous recombination as well as sensitivity to topoisomerase 1 inhibitors (35). Thus, there may be both universal and species-specific mechanisms by which human and mouse WRN maintain genomic integrity.

Molecular and Cellular Endpoints of Genome Maintenance Defects

To understand the basis of accelerated aging that occurs in human and mouse mutants, it is important to consider the molecular and cellular changes that can result from defective genome maintenance. As shown in Fig. 1, a direct consequence of unrepaired DNA damage is transcription interference, which is very likely responsible for the symptoms of TTD and CSB. A general decline in transcription occurs in a variety of organs and tissues with increasing age (41). Moreover, an indirect endpoint of DNA damage is mutation, as a consequence of error-prone repair or misreplication. The accumulation of mutations in organs and tissues during aging could cause the well-known increase in cancer (42) and possibly lead to a variety of functional decrements, especially when large genome rearrangements are involved (43).

Figure 1

Possible molecular and cellular endpoints of aging as driven by DNA damage from energy metabolism. This concept is based on the trade-off between somatic maintenance and growth and reproduction (52).

Apart from the immediate molecular consequences of transcription interference and mutation accumulation, DNA damage may also elicit cellular responses. In mammals, two major cellular responses to genotoxic stress are apoptosis (programmed cell death) (44) and cellular senescence (45). Both of these responses protect against cancer by eliminating severely damaged cells that are at risk for malignant transformation (45). However, while offering such a benefit at younger age, increased loss of functional cells in organs and tissues may explain some prominent symptoms of aging, such as neurodegenerative diseases, loss of immune function, and heart and kidney functional decline.

Of these two cellular endpoints, cellular senescence is least understood. Cellular senescence was first described as replicative senescence, which refers to the limited proliferative potential and eventual growth arrest shown by all normal human and animal cells in culture. The process is often considered as a model for cellular dysfunction in vivo (45). Recent results showing that conditions of high oxygen tension in the culture medium greatly shortens the replicative life-span, particularly of mouse cells, suggests that the senescence response to ROS may underlie differences in aging rates between mice and humans (46). Replicative senescence in humans is caused by telomere shortening, which, if senescence is bypassed through the inactivation of p53 and Rb, results in excessive chromosomal instability. However, senescence in culture also results from treatment with DNA-damaging agents, such as ionizing radiation, mitomycin c, and oxidative stress, as well as through activated oncogenes. Thus, DNA damage can elicit a transient growth arrest, apoptosis, or cellular senescence, all designed to prevent cancer, but apoptosis and cellular senescence may also contribute to aging.

Cells from humans or mice with defects in genome maintenance often undergo accelerated senescence, including cells with genetic alterations described above: human and mouse WRN (35), mouse Ku80 (22), mouse ERCC1 (28), and human and mouse cells with defective telomeres (47). Because senescent cells arrest growth without dying, and acquire altered functions that can in principle disrupt tissue homeostasis, it has been proposed that their accumulation contributes to certain aging phenotypes, including cancer (45, 48). Thus, the direct cause of premature aging seen in repair-deficient mice and humans may not be DNA damage, somatic mutation, or interference with transcription or replication, but rather a cellular response.

The idea that cellular responses may be important for aging is supported by studies of p53, a well-known tumor suppressor protein whose functions are critical for both the apoptotic and senescence responses to DNA damage. Inactivation of p53 rescues some aging phenotypes in mtr –/– mice (47) but accelerates tumorigenesis. Furthermore, deletion of p53 in ku80 –/– mice (17) and wrn –/– mice (49) increases their cancer incidence. p53 inactivation inku80 –/– cells rescues them from premature replicative senescence, but not from hypersensitivity to DNA damaging agents or the ensuing genomic instability (22). Thus, the p53-mediated response, not the accumulation of DNA damage per se, was very likely responsible for the premature senescence and for suppressing cancer.

In keeping with the idea that p53 prevents cancer at the cost of aging phenotypes are findings with mice that express a specific p53 mutant allele, which produces an N-terminally deleted protein (50). The truncated p53 appears to interact with the wild-type p53 protein, which could either have altered function or be regulated differently. As suggested by the authors, this could result in increased p53 activity. Although the homozygous mutant was found to behave like a p53 null mouse, the presence of one mutant and one wild-type allele appeared to cause the premature development of several aging phenotypes, including osteopenia, organ atrophy, diminished stress tolerance, and shortened life-span, while nonetheless reducing cancer. An intriguing possibility is that these aging phenotypes are driven by inappropriate apoptosis or cellular senescence caused by constitutively hyperactive p53. On the other hand, increasing p53 activity through an extra copy of the intact p53 gene, including much of its upstream regulatory region, also reduces cancer incidence in mice but does not influence aging (51).


If defects in genome maintenance lead to accelerated aging in humans and mice, is it possible that normal aging is caused in part by inadequately repaired DNA damage? Genome maintenance is necessarily imperfect, because organisms are not designed to last forever and genetic changes are needed for natural selection. Mutations in one or a few of the possibly hundreds of genes involved in genome maintenance may increase its imperfect nature, possibly in a tissue-specific manner, but might never completely mimic normal aging. Hence, the accelerated aging symptoms in humans and mice with genetic defects in genome maintenance strongly suggest that genome instability, driven by oxidative damage, is a primary cause of normal aging. Genotype-phenotype correlations in mouse models of defective genome maintenance can provide valuable insights into basic mechanisms of aging and the natural defense systems that promote longevity. Such studies can also provide the groundwork for engineering improved DNA maintenance pathways. In turn, this will facilitate genetic or pharmacological interventions that reduce DNA damage, promote repair, or optimize the cellular responses to DNA damage to prolong healthy life (“health-span”) and delay aging.

  • * To whom correspondence should be addressed. E-mail: hastye{at}uthscsa.edu, vijg{at}uthscsa.edu


View Abstract

Navigate This Article