Portal de Periódicos da CAPES

Rebuttal to Miller: ‘Accelerated aging’: a primrose path to insight?’

2004; Wiley; Volume: 3; Issue: 2 Linguagem: Inglês

10.1111/j.1474-9728.2004.00087.x

1474-9726

Paul Hasty, Jan Vijg,

Adipose Tissue and Metabolism

In his ‘Accelerated aging’: a primrose path to insight?’ Richard Miller discusses problems and pitfalls associated with the use of animal models of accelerated aging for obtaining insight into mechanisms of longevity and aging. The issues raised, some of which we addressed in our paper ‘Accelerating aging by mouse reverse genetics: a rational approach to understand longevity’, are valid and deserve attention. Most significantly, we agree that side-by-side comparisons of aging-related phenotypes in mutant and control mice are essential. In addition, it is desirable that multiple tissues be affected and that the premature appearance of aging phenotypes occurs after development, preferably after maturation. Why then do we disagree with Miller about the potential usefulness of accelerated mouse models in elucidating normal aging? We believe there are four basic points of disagreement. The first three points involve interpretation of data. The fourth point is theoretical and influenced by the first three. The first point of disagreement is that Richard Miller believes that individuals exhibiting accelerated senescence do not look like normally aged individuals. He makes this point in two figures that show progeroid humans and mice not resembling wild-type aged individuals. However, his comparisons are misleading. In Figure 1 he compares aged human adults with a child with Hutchinson–Gilford progeria (mutation in A-type lamin). As we mention in our paper, the most convincing accelerated aging models display age-related phenotypes only after maturation. Naturally, a child who ages before maturation will look unusual as compared with an aged, fully matured person. Werner's syndrome is a better example for making this kind of comparison. People with Werner's syndrome show signs of accelerated aging only after maturation and are indistinguishable in appearance from control-aged individuals. In his Figure 2, Miller compares wild-type mice with those with a dominant-mutation in p53 (the p53+/m mice) or a mutation in the Klotho gene. The Klotho mutant mice also age before maturation, similar to humans with Hutchinson–Gilford, and they therefore look unusual. However, the p53+/m mice exhibit aging characteristics after maturation. So why do they not look like control aged mice? In fact they do when the correct comparison is made (L. Donehower, personal communication). This is also true for the Ku80-mutant mice as shown in Figure 3 of our paper. We believe the discrepancy between observations by Donehower and ourselves with those by Richard Miller is due to differences in genetic background. A true comparison in mice can only be made between littermates raised under identical conditions. The importance of genetic background for phenotype is best illustrated by mutation of the epidermal growth factor receptor (EGFR) gene (Threadgill et al., 1995). Deletion of EGFR results in peri-implantation death due to degeneration of the inner cell mass in a CF-1 background, mid-gestation death due to placental defects in a 129/Sv background or postnatal death with abnormalities in skin, kidney, brain, liver and gastrointestinal tract in a CD-1 background. Therefore, analysis of the EGFR mutation in a single background elucidates only a segment of the potential phenotypes. Genetic background would probably be an even greater consideration for aging because, unlike development, aging is not highly regulated. The difference in cancer incidence and spectrum is a good example of an age-related phenotype that varies with mouse strain (Tripodis et al., 2001). We believe that isogenic mouse strains exhibit only a segment of the possible aging phenotypes and as a result comparing one strain with another could be misleading. The second point of disagreement is that for human and mouse progeroid syndromes, only a fraction of all possible age-related characteristics are prematurely expressed; as a consequence these models have been termed ‘segmental’. However, as we discuss in our paper, every normally aged individual exhibits a segmental phenotype when compared with all possible phenotypes in the population. Segmental aging is natural and to be expected, and is probably due to, in part, genetic variation in the population. Moreover, the most convincing human or mouse models of accelerated aging exhibit a wide range of aging-related phenotypes, when compared with the appropriate control individuals. Such pleiotropy of a heritable mutation in a candidate longevity assurance gene is difficult to explain as mere coincidence and the results obtained with mouse random mutagenesis projects essentially rule out the possibility that hundreds of different random mutations can produce complex phenotypes resembling aging. Again, valid comparisons can only be made in the same genetic background. Indeed, the label of segmental aging is unfairly applied when comparisons are made between different backgrounds. The third point of disagreement is Miller's argument that DNA repair is unlikely to be important for aging because not all DNA-repair-mutant mice exhibit accelerated aging. In our opinion, this argument ignores the complexity of DNA repair and genome maintenance. Indeed, over 100 genes are involved in DNA repair per se (Wood et al., 2001) and many more in genome maintenance overall. It is certainly conceivable that in human and mouse populations, polymorphic variation at loci of genome maintenance could act as determinants of segmental aging (Vijg & van Orsouw, 2002). However, mutations at some of these loci could cause a more severe imbalance than mutations at other loci. In addition, deletion of some proteins is embryonic-lethal or causes cancer at an early age, preventing additional aging phenotypes to become manifest. Therefore, a subtle mutation or an additional defect, such as decreased cell-cycle response, is often required to visualize the phenotype, as for example in the case of the Xpd and Brca1 mutant mice, respectively (Hasty et al., 2003). Therefore, only certain fortuitous mutations, i.e. relatively benign – so as to become manifest only after maturation – and occurring in genome maintenance genes that are important across multiple tissues, will cause such dramatic accelerated aging phenotypes as in Werner's syndrome. The fourth point of disagreement involves a different perspective on the fundamental causes of aging. Indeed, the real difference between Richard Miller and us seems to involve our respective ideas about what aging is and how it might work. Richard Miller refers to this issue frequently and believes the scientific community does not yet have a sound idea of what causes aging. However, in our opinion there is such an idea and it is based on damage accumulation. With his disposable soma theory, Kirkwood (1977) has provided a perfectly logical argument to predict that aging results from a gradual accumulation of damage in cells and tissues. Damage accumulation as a cause of aging is not in conflict with certain apparent programmatic characteristics of aging. However, rather than programmed to age, organisms are programmed for survival, through the activity of a network of cell-maintenance and stress-response systems. As we freely distill from Richard Miller's manuscript and recent work, we believe his view on aging is sympathetic to that of a mechanistically synchronized series of time-dependent processes, controlled by sets of master genes. This viewpoint is highly popular among those attempting to explain aging in terms of few relatively simple, universal mechanisms, originally based on findings of single mutations extending lifespan of the nematode (Guarente & Kenyon, 2000). In our opinion, there is no factual basis for universal aging pathways. Instead, what these studies have consistently shown is that increased somatic maintenance and stress resistance are associated with increased lifespan. They do not provide evidence for a universal timing mechanism for the myriad of aging phenotypes in any other way than through the accumulation of damage. Indeed, plasticity of lifespan as a function of different circumstances in nematodes, fruit flies and mice can best be explained in terms of metabolic switches, affecting levels of spontaneous somatic damage (e.g. reactive oxygen species) and/or proportional effort devoted to somatic maintenance (Kirkwood & Austad, 2000). We believe that a major component of the stochastic basis of aging is the random accumulation of DNA damage and mutations, which increasingly hinders cellular functions and stimulates responses resulting in cellular senescence or apoptosis (Hasty et al., 2003). Segmental progeroid phenotypes in human and mouse mutants with heritable defects in genome maintenance pathways offer strong support for this idea, without ruling out additional genes that may be significant for species-specific age-related health problems, such as amyloid precursor protein and presenilin in Alzheimer's disease (Hardy, 1997). Some heterogeneity in the type of genes with late-age adverse effects is predicted by the ‘mutation accumulation theory’ (Kirkwood & Austad, 2000). A general recognition of the validity of mouse models of accelerated aging is important in view of their application in studying specific aging phenotypes and the development of interventions to postpone or prevent them. Normal mice are impractical as a model for studying specific aging phenotypes due to incomplete penetrance and late onset. Analogous to mouse models for human cancer, short-term mouse models for aging, based on molecular defects in a basic longevity assurance mechanism, offer unique opportunities for testing interventions.