Online Issues
<< All Back-issues
<< This Issue's Table of Contents
ILAR Journal V38(3) 1997
Animal Models of Aging Research
Richard L. Sprott and Israel Ramirez
| Richard L. Sprott, Ph.D., is Associate Director at the National Institute on Aging, National Institutes of Health, Bethesda, Maryland. Israel Ramirez, Ph.D., is Adjunct Professor at Rosemont College, Rosemont, Pennsylvania. |
INTRODUCTION
Animal models are commonly used in aging research because they allow researchers to obtain data that are difficult or impossible to obtain from humans. Some studies that can, in principle, be done in humans are much easier to do in animals because of their shorter life spans. For example, measurements of changes over entire life spans, which would take many decades to complete in humans, can be accomplished in rodents in 2 to 3 years. Other kinds of experiments cannot, even in principle, be done with humans. Using rodents, researchers can manipulate breeding, extract tissues, administer bioactive agents, control diets, and perform surgical manipulations--procedures that would be unethical or excessively hazardous using humans.
Rats and mice are the species of choice for most animal aging studies for a number of reasons: Techniques for breeding and housing them are well developed; they age rapidly; and a very large body of knowledge about their biology and behavior exists, which provides a solid base for future advances. Rats and mice thus provide indispensable tools for the study of mammalian aging and the development of methods for ameliorating aging-related diseases in humans.
The scientific issue is not whether to study rats and mice but instead, which rats and mice should be studied. The discussion that follows reviews the factors that determine the choice of strain and species. This discussion is limited to mice and rats; other species are discussed by Sprott and Austad (1996).
MODEL AVAILABILITY AND COST
A substantial investment in husbandry must be made before aged rodents can be used in an experiment. They must be housed, fed, and maintained for 3 or more years. The rodents must be maintained in barrier facilities to ensure that they do not acquire infectious diseases, and their health status must be regularly surveyed. Furthermore, a substantial fraction of the rodents will have died before reaching the desired age, making it necessary to maintain as many as 3 rodents for every 1 used in an experiment at ages ranging from 24 to 36 months. As a result, aged rodents can be the single most expensive item in an investigator's budget. Anything that reduces animal costs--without compromising the quality of the research--increases the amount of research an investigator can conduct.
Reducing interanimal and interlaboratory variability permits the use of fewer animals and thereby reduces costs. For this reason, the National Institute on Aging (NIA) has issued statements encouraging investigators to use inbred strains and F1 hybrids. Many generations of inbreeding (brother-sister mating) minimizes genetic variability and produces animals that are essentially homozygous at all genetic loci. The absence of genetic variability can be desirable because it is a major source of variation between individuals and between experiments. Homozygosity, however, is less desirable because it is an abnormal condition. Many deleterious genes are recessive, that is, their effects are manifest only when the individual is homozygous for that gene. During inbreeding, some deleterious genes are lost through natural selection, but others are fixed, resulting in less fit animals.
Crosses between inbred strains (F1 hybrids) provide the genetic uniformity of inbred strains without the deleterious effects of homozygosity. Furthermore, F1 hybrids are usually less variable than inbred strains (Phelan and Austad 1994). The mechanisms responsible for the reduced variability in F1 hybrids are not fully understood. It is likely that this effect is attributable partly to a decrease in the effects of recessive genes having deleterious effects and partly to biochemical advantages resulting from having two versions of many enzymes (Phelan and Austad 1994). Whatever the mechanism, the crucial point is that one can reduce the number of animals tested, and hence reduce the cost of research, by using FI hybrids.
F1 hybrids offer yet another advantage. They often live longer (Finch 1990), presumably because of their better health. For example, male rats of the Fischer 344 (F344) and Brown Norway strains are reported to have median life spans of 103 and 129 weeks, respectively, whereas their F1 hybrid has a median life span of 145 weeks when fed ad libitum. As a result of reduced mortality, more old animals will live long enough to become available for a study. The availability of rats of 2 inbred strains and their F1 hybrid also makes it possible for an investigator to observe simple genotype/environment interactions.
Many investigators do not have access to large barrier vivaria in which to rear the animals needed for their research. Moreover, it is not economical for the NIA to support separate animal colonies at each of the more than 300 NIA-sup-ported laboratories in the United States where rodents are used in research. For these reasons, the NIA created large colonies of aging mice and rats in barrier facilities, maintained under contracts between NIA and commercial breeders. These contracts supply more than half of all the animal models currently used by NIA grantees in their research. The economies of scale permit NIA to supply high quality rats and mice to investigators at a cost they can afford. NIA also makes small numbers of these animals available at no cost for pilot studies and dissertation research through a simple application process with rapid turnaround time.2
The availability of these rodents has played an important role in the development of gerontological science in the 20 years since the NIA colonies were established. At the same time, this easy availability has promoted reliance on a relatively limited set of mouse and rat genotypes. (Other articles in this issue discuss alternative mouse genotypes (McClearn, "Heterogeneous Reference Populations in Animal Model Research in Aging"; Richardson and others, "Use of Transgenic Mice in Aging Research"; Takeda and others, "Senescence-accelerated Mouse (SAM): With Special Reference to Development and Pathobiological Phenotypes") and other animal species (Austad, "Birds as Models of Aging in Biomedical Research" and "Small Nonhuman Primates as Potential Models of Human Aging"). The use of "designer" mice and alternative species offers significant opportunity for new insights into important aging processes.
GENERALIZABILITY OF RESULTS
The primary reason for studying rats and mice is to obtain insights about gerontological processes that are broadly applicable to mammals overall and humans in particular. Since the strain or species chosen for study might have unusual attributes not shared by other mammals, researchers should avoid focusing on a single strain or species. Several years ago, Weindruch and Masoro (1991) argued that investigators were relying too heavily on F344 rats. This strain was used in a majority of the reports employing rats published in the Journal of Gerontology between 1985 and 1990. Weindruch and Masoro (1991) pointed out that since each strain is more susceptible to some diseases than to others, excessive reliance on a single strain could turn research into the study of the effects of a few diseases, rather than the effects of aging per se. NIA developed the F344 x BNF1 cross specifically to provide an alternative to the heavy use of the F344 rat in research. Overall, NIA has tried to establish the principle that no model is inherently "good" or "bad." Rather, it is imperative that investigators make informed choices of models.
CURRENT RODENT USAGE
To determine whether usage patterns are changing, we tabulated strain usage over the most recent 2 years (mid- 1994 to mid-1996) in the following journals: Mechanisms of Aging and Development and the Journals of Gerontology, Biological Sciences, Experimental Gerontology, Aging: Clinical and Experimental Research, and International Journal of Experimental and Clinical Gerontology. Studies were divided into 3 groups, those from the 2 geographic regions producing the largest number of studies (United States and Europe) and everywhere else. As seen in Table 1, researchers tend to focus on very few genotypes and the choice of genotype depends on geographic region. Accordingly, 64% of rat studies conducted in the United States employed F344 rats whereas 52% of European studies employed one of the Wistar sublines. Studies employing mice show a similar, but less extreme, trend: 61% of mouse studies in the United States employed one of the C57BL/6 sublines. Researchers outside the United States employed a wider variety of mouse strains, with no one strain accounting for more than 30% of the studies. In the United States, at least, the concentration on F344 rats and C57BL/6 mice is not simply a reflection of availability.
As described above, aged animals representing a wide variety of mouse genotypes and 3 rat genotypes are easily available from NIA-sponsored colonies. Rather, it appears that many investigators choose animals based on common usage, as reflected in the scientific literature. One can speculate that this seems "safe" and that investigators choose animals in this way especially if they are not motivated to choose a less well-known model. This behavior, of course, perpetuates the reliance on a very restricted range of models, dramatically increasing the risk that research results have limited generality.
Since publications describe work done 1 or more years ago, we considered the possibility that a recent shift in strain usage would not immediately appear in publications. We obtained information about strain selection from ongoing studies in the United States by examining distribution records from the NIA contractors (Hadan Sprague Dawley and Charles River Laboratories) for July 1995 through June 1996. We found that 74% of the rats shipped were F344, and 59% of mice were C57BL/6.
"Outbred" rat stocks (Wistar and Sprague Dawley) are still used frequently in aging research, especially outside the United States (Table 1). It should be kept in mind that such stocks have not been standardized. Rats of the same outbred stock, but coming from different suppliers, can differ substantially (Klinger and others 1996).
Despite the advantage of F1 hybrids, few of the studies examined used them (Table 1). This appears to be changing, because approximately 19% of rats and 7% of mice shipped by the NIA contractor in the most recent 12-month period were F1 hybrids.
Animal usage is also restricted in the employment of more male than female rats, especially in those studies conducted in the United States (Table 2). Recent data on shipments of NIA rodents show an even greater reliance on males: 88% of rats and 67% of mice were male.
The reasons for the preferential use of males are varied. The most commonly cited reason is that variability due to estrous cycles could complicate the work. However, if this variability is actually significant, it should promote more, rather than less, usage. In other words, if female hormone cyclicity has significant effects on the phenomena being investigated, then that same variability could offer valuable clues to the underlying biological processes. Does it really make sense to discard this potentially valuable information?
Another obvious reason for using animals of only one sex is cost economy. Using male and female animals increases the number of animals needed for a study. Here again, however, the economy may be false. In the short run, single sex studies may be cheaper, but what happens if the results of the study are interesting? As long as the phenomenon is not clearly sex-related (such as prostate or endome-trial cancers), studies that produce interesting results (positive or negative) will have to be repeated using the sex not represented in the first study. Repeating the study, perhaps even several times, will be more expensive than an initial comprehensive study would have been. Neither mice nor rats produce offspring in a 2: or 3:1 ratio of males to females. Reliance on sex results in the waste of many animals of the "wrong" sex.
An additional disadvantage of the exclusive use of males is that they grow at a more rapid rate, for a longer period of their lives, than females (Sprott and Austad 1996). An experiment comparing males less than 1 year of age and males 2 years of age or older is basically a comparison of rapidly growing versus static animals unless food intake is restricted. Complications resulting from high growth rates in young animals can be mitigated by using females.
DISEASE-RELATED COMPLICATIONS
Disease incidence increases with aging. Very old animals ordinarily suffer from one or more diseases, making it difficult to separate the effects of aging from the effects of various diseases. This task is further complicated by the tendency of each strain to suffer from characteristic diseases (Bronson 1990; Lipman and others 1996). F344 rats, for example, commonly suffer from nephropathy and testicular tumors (Weindruch and Masoro 1991). Although age-related diseases are unavoidable, it is possible to minimize their adverse effects on research by checking the literature before initiating a study to determine the diseases to which particular strains are prone. For example, before initiating research, an investigator interested in aging in the kidney should identify a strain that is not susceptible to nephropathy.
The most important consideration in the choice of an animal model for any research is suitability. It is the responsibility of every investigator to become knowledgeable about the animal models chosen for use in a research program. Failure to acquire basic information can lead to errors such as the unwitting use of blind mice in studies of visual maze performance or melatonin-deficient mice in studies of the role of melatonin in longevity.
COMPARISONS BETWEEN GENOTYPES
Researchers can do more than merely avoid strains that could invalidate the results. The existence of various genotypes makes it possible to choose those with especially desirable attributes or to compare genotypes to gain insights about mechanisms.
An obvious way to exploit strain differences is to compare strains that differ in life span. For example, Gilad and Gilad's (1995) review suggests that longevity of various strains is correlated with their neuroendocrine stress responses. The problem with this approach is lack of certainty as to whether such interstrain variation is due to differences in the rate of aging or susceptibility to disease (Finch 1990; Harrison 1994). This issue is not easy to resolve because there is no universally accepted way to discriminate between the 2 possibilities. One approach is to examine the mathematical relationship between age and mortality. One widely used technique is to partition mortality into age-independent and age-dependent components via the Gompertz equation (Finch 1990). The age-indepen-dent component provides a measure of overall disease susceptibility (or environmentally induced disease), and the age-de-pendent component provides a measure of the rapidity with which a strain or species ages (Finch 1990). Comparisons of the mortality rate based on the Gompertz model suggest that susceptibility to disease is responsible for most strain differences; the age-independent mortality rate differs between strains to a greater degree than the age-dependent acceleration of mortality (Finch 1990; Finch and Pike 1996). The F344 X Brown-Norway Fl hybrid might be an exception to this generalization since it shows delayed onset of accelerated mortality compared with the F344 strain (Finch and Pike 1996; Sprott and Austad 1996).
Senescence-accelerated mouse (SAM) strains have been developed to provide a model of accelerated aging. These mouse strains, derived from a cross between the AKR/J strains and another unknown progenitor, differ greatly in life span (Takeda 1994). Strains designated as senescence-prone acquire an aged appearance, accumulate amyloid in many tissues, and die in less than a year. Whether or not one accepts the conclusion that these strains show accelerated senescence, there is no question that they show a remarkable number of symptoms that resemble ordinary aging including decline in immune function, osteoporosis, degenerative joint disease, cataracts, hearing loss, senile-hyperinflation of lung, decreased ability to learn, and disability (Takeda 1994). For a full description of the SAM mice, see Takeda and others, "Senes-cence-accelerated Mouse (SAM): With Special Reference to Development and Pathological Phenotypes," in this issue.
When comparing strains, it is important to remember that they may differ in many traits as the result of accidents of inheritance. Evidence that long- and short-lived strains differ in some measure does not establish a causal relationship between the measure and longevity. Strain comparisons can be strengthened by examining crosses between the strains (FI hybrids). Evidence that the traits of interest show the expected relationship in F1 hybrids strengthens evidence for a causal relationship. By breeding the FI hybrids, one can obtain a well-defined but genetically variable population (F2 generation), which can be used to critically test correlations between traits among individuals. Brother-sister mating of the animals from the F2 generation for 20 or more generations produces new, recombinant inbred strains. These new strains vary among each other and hence can be used to test correlations. They have an important advantage over F2 animals in that individuals from a given recombinant inbred strain can be reproduced. Such strains have enabled the identification of linkage groups associated with long and short life spans (Gelman and others 1988). Additional discussions of the power of recombinant inbred strains and other techniques for genetic analysis of aging have been published elsewhere (Green 1981; Sprott 1993).
The existence of strains with retarded aging has not been conclusively established. An alternative technique for obtaining senescence-retarded animals is to restrict their caloric intake over a large portion of their lives (Masoro 1992; Sprott and Austad 1996). This technique has been shown to reduce the age-related component of mortality (Finch 1990; Finch and Pike 1996) as well as to reduce the incidence of diverse age-related diseases (Masoro 1992; Finch 1990; Nelson and others 1995; Yu 1994). Such animals are being provided by the National Institutes on Aging.
AGES EXAMINED
Some measures that change monotonically with age (such as histologic lesion incidence (Lipman and others 1996)) could serve as biomarkers for aging. Other measures change in a nonmonotonic fashion; for example, body weight first increases and then usually decreases with age (Sprott and Austad 1996). This is not particularly surprising but has interesting implications for the choice of animals for research.
In the course of surveying the literature summarized in Tables 1 and 2, it became apparent that half the publications examining aging employed only 2 age groups (Table 3). An additional 20% of publications employed only 3 age groups (see Table 3). When only 2 ages are examined, it is not possible to determine whether any observed age effects are due to a change that occurs throughout life or merely over some restricted portion of life. For any variable that first increases and then decreases with age (such as body weight), the outcome of a study employing only 2 ages will be critically dependent on which 2 ages are examined. Even a study employing 3 age groups has a very limited ability to establish how a single variable changes with age. Several N1A-sup-ported genotypes (mouse and rat) live 30 months or more when fed ad libitum and 36 months or more when led a calorically restricted diet (Sprott and Austad 1996). Studies using at least 3 age groups might include young adults (8 to 12 months), middle-aged (18 to 24 months), and old (30 to 36 months) subjects.
CONCLUSION
The choice of animal models for research is one of the most critical aspects of experimental design. Far too often it is done with too little information. Choices are made on the basis of convenience (such as easy access from a commercial vendor), a desire to avoid offending grant reviewers, or worse yet, simple ignorance. Fortunately, some aspects of this situation are improving. Investigators are becoming more knowledgeable. Publication of good research using a wider variety of animals results in more such research. Furthermore, the development of genetically engineered animal models (see Richardson and others, "Use of Transgenic Mice in Aging Research," this issue) is dramatically increasing general awareness of the contribution of genotype to suitability of particular animal models for specific types of research. This increased awareness will likely lead to a generalized improvement in gerontological research as a whole. Similarly, the use of additional species (see Austad, "Birds as Models of Aging in Biomedical Research" and "Small Nonhuman Primates as Potential Models of Human Aging," both in this issue) can be expected to contribute to an increased awareness of the limitations of reliance on closely related species for almost all gerontological research.
1Abbreviations used in this article: NIA, National Institute on Aging; SAM, senescence-accelerated mouse.
2The NIA colonies currently include the following rodents:
Calorically restricted rats: F344, BN/BiRij; mice: C57BL/6NNia, B6D2F1NNia.
Ad libitum-fed rats: F344, F344 (retired breeders), BN/BiRijNia, F344BNFI/Nia; mice: BALB/cByNia, CBA/JNia, C57BL/6JNia, DBA/ 2JNia, B6C3FI (C57BL/6 x C3H), B6D2F1 (C57BL/6 x DBA2), CB6FI (BALB/C x C57BL6), CRL:COBS Swiss Webster Outbred.
For information and application forms, contact DeWitt Hazzard, Ph.D., NIA/NIH/BAP, Gateway Building, Room 2C231, 7201 Wisconsin Avenue. Bethesda, MD 20892.
REFERENCES
Bronson RT. 1990. Rate of occurrence of lesions in 20 inbred and hybrid genotypes of rats and mice sacrificed at 6-month intervals during the first years of life. In: Harrison DE, editor. Genetic Effects on Aging Il. Caidwell NJ: The Telford Press. p 279-358.
Finch CE 1990. Longevity, Senescence, and the Genome. Chicago: University of Chicago Press.
Finch CE, Pike MC. 1996. Maximum life span predictions from the Gompenz mortality model. J Gerontol Biol Sci 51A:BI83-BI94.
Gelman, R, Watson A, Bronson R, Yunis E. 1988. Murine chromosomal regions correlated with longevity. Genetics 118:693-704.
Gilad GM, Gilad VH. 1995. Strain, stress, neurodegcncration and longevity. Mech Ageing Dev 78:75-83.
Green EL. 1981. Genetic methods in animal experimentation. In: Gay WI, editor. Methods of Animal Experimentation. Vol VI. New York: Academic Press.
Harrison DE. 1994. Potential misinterpretations using models of accelerated aging. J Gerontol Biol Sci 49:B245.
Klinger MM, MacCarter GD, Boozer CN. 1996. Body weight and composition in the Sprague Dawley rat: Comparison of three outbred sources. Lab Anita Sci 46:67-70.
Lipman RD. Chrisp CE, Hazzard DG, Bronson RT. 1996. Pathologic characterization of Brown Norway, Brown Norway X Fischer 344, and Fischer 344 X Brown Norway rats with relation to age. J Gerontol Biol Sci 51A:B54-B59.
Masoro EJ. 1992. A dietary key to uncovering aging processes. News Physiol Sci 7:15-160.
Nelson JF, Karelus K, Bergman MD, Fellcio LS. 1995. Neuroendocrinc involvemere in aging: Evidence from studies of reproductive aging and caloric restriction. Neurobiol Aging 16:837-843.
Phelan JP, Austad SN. 1994. Selecting animal models of human aging: Inbred strains often exhibit less biological uniformity than Fl hybrids. J Gerontol 49:BI-BII.
Sprott RL. 1993. Mouse and rat genotype choices. Aging Clin Exp Res 5:249-252.
Sprott RL, Austad SN. 1996. Animal models for aging research. In: Schneider EL, Rowe JW, editors. Handbook of the Biology of Aging. 4th ed. Orlando: Academic Press. p 3-24.
Takeda T, editor. 1994. The SAM Model of Senescence. Amsterdam:Excerpta Medica.
Weindruch R, Masoro EJ. 1991. Concerns about rodent models for aging research. J Gerontol Biol Sci 46:B87-B88.
Yu BP. 1994. How diet influences the aging process of the rat. Proc Soc Exp Biol Med 205:97-105.
| Fischer 344 | Wistar | Sprague Dawley | Brown Norway | F1 | Other | C57BL/6 | BALB/ C | SAM | CBA/2 | F1 | Other | |
| United States | ||||||||||||
| N | 37 | 3 | 9 | 1 | 4 | 4 | 11 | 5 | 0 | 1 | 0 | 1 |
| % | 64 | 5 | 16 | 2 | 7 | 7 | 61 | 28 | 0 | 6 | 0 | 6 |
| Europe | ||||||||||||
| N | 2 | 22 | 12 | 2 | 0 | 4 | 3 | 2 | 0 | 1 | 0 | 4 |
| % | 5 | 52 | 29 | 5 | 0 | 10 | 30 | 20 | 0 | 10 | 0 | 40 |
| Elsewhere | ||||||||||||
| N | 5 | 5 | 6 | 0 | 0 | 1 | 3 | 2 | 5 | 5 | 1 | 8 |
| % | 29 | 29 | 35 | 0 | 0 | 6 | 15 | 10 | 25 | 10 | 5 | 35 |
| Total | ||||||||||||
| N | 44 | 30 | 27 | 3 | 4 | 9 | 17 | 9 | 5 | 4 | 1 | 12 |
| % | 38 | 26 | 23 | 3 | 3 | 8 | 35 | 19 | 10 | 8 | 2 | 25 |
N represents the number of studies using each genotype during the most recent 2 years (mid-1994 to mid-1996) in the following journals: Mechanisms of Ageing and Development; The Journals of Gerontology: Biological Sciences; Experimental Gerontology; Aging: Clinical and Experimental Research; and International Journal of Experimental and Clinical Gerontology. Percentage values are the percentage of the total number of studies (not animals) in each geographic region. Rat and mouse percentages are computed separately.
Closely related strains were grouped together if they are commonly referred to by the same name (for example, Sprague Dawley).
Percentages have been rounded and do not always total 100.
TABLE 2 Gender of rodents used
| Male | Female | Both | |
| United States | |||
| N | 57 | 11 | 4 |
| % | 79 | 15 | 6 |
| Europe | |||
| N | 31 | 9 | 6 |
| % | 67 | 20 | 13 |
| Elsewhere | |||
| N | 12 | 5 | 4 |
| % | 57 | 24 | 19 |
| Total | |||
| N | 100 | 25 | 14 |
| % | 72 | 18 | 10 |
Data are similar to Table 1 except that gender of the animals used is given instead of genotype. Data from rats and mice have been collapsed. Studies that did not clearly indicate gender of the animals are not included in this table.
TABLE 3 Number of age groups compared
| 2 | 3 | More | |
| United States | |||
| N | 43 | 17 | 18 |
| % | 55 | 22 | 23 |
| Europe | |||
| N | 23 | 12 | 18 |
| % | 43 | 23 | 34 |
| Elsewhere | |||
| N | 15 | 4 | 11 |
| % | 50 | 13 | 37 |
| Total | |||
| N | 81 | 33 | 47 |
| % | 50 | 20 | 29 |
Data are similar to Table 1 except that the number of age groups in each publication is given.
Percentages have been rounded and do not always total 100.
Copyright © 2008. National Academy of Sciences.
All rights reserved.
500 Fifth St. N.W., Washington, D.C. 20001.
Terms of Use and Privacy Statement