Bring Back the Rat!

Abstract

As 2020 is “The Year of the Rat” in the Chinese astrological calendar, it seems an appropriate time to consider whether we should bring back the laboratory rat to front-and-center in research on the basic biology of mammalian aging. Beginning in the 1970s, aging research with rats became common, peaking in 1992 but then declined dramatically by 2018 as the mouse became preeminent. The purpose of this review is to highlight some of the historical contributions as well as current advantages of the rat as a mammalian model of human aging, because we suspect at least a generation of researchers is no longer aware of this history or these advantages. Herein, we compare and contrast the mouse and rat in the context of several biological domains relevant to their use as appropriate models of aging: phylogeny/domestication, longevity interventions, pathology/physiology, and behavior/cognition. It is not the goal of this review to give a complete characterization of the differences between mice and rats, but to provide important examples of why using rats as well as mice is important to advance our understanding of the biology of aging.

As 2020 is “The Year of the Rat” in the Chinese astrological calendar, it seems an appropriate time to consider whether we should bring back the laboratory rat to front-and-center in research on the basic biology of mammalian aging. One might ask whether the rat has ever in fact disappeared from aging research? Not wholly, we would answer, but its prominence has been eclipsed in recent years by the laboratory mouse. To illustrate this point, we compared the number of papers returned in a simple search of Medline for the terms “mouse” combined with “aging” versus “rat” combined with “aging” (Figure 1, bottom). As can be seen, in the 1960s, rat and mouse papers on aging were published at approximately at the same rate. Beginning in the 1970s, aging research with rats became more common, peaking in 1992 when the ratio of rat-to-mouse papers reached 3.0, a number that by 2018 had nearly been reversed. This general pattern was also reflected in discipline-specific journals such as the Journal of Gerontology from 1946 to 1994, which subsequently became the Journals of Gerontology: Biological Sciences from 1995 (Figure 1, top).

Figure 1.

Demonstrates the rise of the mouse in aging research and the decline of the rat. The top figure reflects number of papers with rats as subjects in discipline-specific journals: the Journal of Gerontology from 1946 to 1994, which subsequently became the Journals of Gerontology: Biological Sciences from 1995. The bottom figure reflects number of papers returned in a simple search of Medline for the terms “mouse” combined with “aging” versus “rat” combined with “aging”.

The reason for this shift is clear. The mouse rose to prominence in nearly all fields of biomedical research during the late 1980s and early 1990s as our ability to modify individual genetic loci in the mouse genome emerged (1). This shift also occurred in aging research as highlighted by the 15-year-long National Institute of Aging (NIA) program (the Longevity Assurance Gene [LAG] Program), initiated in 1989 to search for genes associated with increased longevity. This program took a single gene manipulation approach to identify basic pathways of aging, by translating findings from invertebrate model organisms, especially worms and flies, to the newly tractable mouse genome (2). Because of this genetic tractability, use of the mouse began to replace that of the rat, which never had a place in the LAG program, a process that was reinforced by the priority given to the sequencing of the entire C57BL/6J mouse genome in 2002 (3) and follow-up studies refining the sequence and extending this approach to other mouse genotypes.

Once a number of these homologous genetic pathways were identified across worm, fly, and mouse species, pharmacological targets were sought for translational potential, ultimately leading to the development of the Interventions Testing Program (ITP), ongoing since 2004, to identify compounds that increased mouse longevity. The ITP consists of multisite interventions with exceptional standardization among sites to maximize rigor and reducibility. One particularly surprising finding of the ITP has been that a number of mouse life-extending drugs had sex-specific or sex-biased effects. This program still exists and serves as the cornerstone of validating longevity interventions in mice.

So what is the rationale for bringing the rat back to prominence? The purpose of this review is to highlight some advantages of the rat as a mammalian model of human aging, because we suspect at least a generation of researchers may no longer be aware of these advantages. Some obvious reasons are size-related. Rats are larger, thus have more tissue for biochemical/physiological/pathological analyses. Larger organs increase the ease of surgical procedures and dissection of relatively smaller anatomical structures. In addition, CRISPR/Cas technology means that creating knockout rat models no longer present a significant technical impediment, obviating one major experimental advantage of mice. Additionally, there is a rich literature within in aging research describing longevity patterns as well as the physiological effects of interventions, and their impact on pathology in commonly used rat models. Furthermore, rats, compared with mice, are ideal subjects for studying functional outcomes as rats are more social, less anxious, and easy to handle, and there is nearly 50 years of behavioral neuroscience–based research using rats as subjects in terms of translating functional/cognitive/behavioral outcomes. In fact, most behavioral tasks utilized in mice today are based upon research in rats and were designed to take advantage of the natural ecological behavior of rats. In addition, the rat is the chosen translational preclinical model for the Common Fund supported The Molecular Transducers of Physical Activity Consortium (MoTrPAC). Below, we compare and contrast the mouse and rat in the context of several biological domains relevant to their use as appropriate models of aging: phylogeny/domestication, longevity interventions, pathology/physiology, and behavior/cognition. It is not the goal of this review to give a complete characterization of the differences between mice and rats, but to provide important examples of why using rats as well as mice is important to advance our understanding of the biology of aging.

Origin and Diversity

The laboratory rat is a domesticated version of the Brown rat (Rattus norvegicus). Despite the species name (norvegicus) and alternative common names, the Norway or Brown Norway rat, ancestors of the laboratory rat in fact originated in east Asia, where they thrived in association with human habitations. Rats spread to Europe along trade routes possibly as late as the 1500s and from there to the New World in the 1700s (4,5).

Within rodents, laboratory rats and mice (Mus musculus) diverged from their common ancestor only 12–24 million years ago (6). To put this in perspective, within primates, humans and African monkeys, such as macaques and baboons, diverged from one another approximately 25 million years ago (7,8). Moreover, ~90% of rat genes have strict orthologs in both human and mouse genomes (6).

Domesticated brown rats have been regularly used in physiological and behavioral research in Europe since the 1850s and in the United States since the 1890s (9). One of the key places for the introduction of rats into research in the United States was Philadelphia’s Wistar Institute, founded in 1892, although a number of other breeding and commercial production facilities subsequently arose.

A key difference between laboratory mice and rats—one that we feel deserves special attention—is in the genetic diversity represented in commonly used domesticated stocks and strains. Virtually all of the traditional inbred laboratory strains of mice originated from a small number of “fancy mice,” so their diversity is highly constrained (1). In contrast, laboratory rats have come from multiple domestications and in fact were often bred back to wild-caught animals. For instance, the Brown Norway (BN) rat strain of today was created from wild rats caught in and around Philadelphia in the second decade of the 20th century. Around the same time on the other side of the country, the Long-Evans rat strain originated from the breeding of an albino Wistar female with a wild-caught California male rat (9). As a consequence of this chaotic domestication history, many of the major rat stocks and strains used in biomedical research such as Wistar-derived, Sprague-Dawley (SD)-derived, and BN strain have mitochondrial genomes from distinct clades which diverged a minimum of 13 thousand years ago (10). The origin of several of the rat strains most commonly used in aging research is given in Table 1 (9).

Table 1.

Origin of Some of the Commonly Used Rat Genotypes in Aging

Genotype	Origin	Inbred/ Outbred
Brown Norway (BN)	4 wild-caught rats (Philadelphia) ~1930	Inbred
Fischer 344 (F344)	Unknown. Inbred since 1920.	Inbred
Sprague-Dawley	Breeding between hooded male and female Wistar ~1925	Outbred
Wistar	Obscure. Origin either Europe or United States. Used in United States from 1906	Outbred
Long-Evans	Wild-caught gray male and Wistar white females ~ 1915	Outbred

Today, there exist more than 700 inbred strains and 60 outbred stocks of rats along with nearly 250 transgenic strains (11). The BN reference strain genome sequence was published in 2004 (6), and there are now complete genomes available for more than two dozen other strains from which more than 80 million variants have been identified (11). Extensive genomic and phenotypic information is available through the Rat Genome Database (RGD, http://rgd.mcw.edu), which also provides many analytic tools for mining these genomes. The Rat Resource and Research Center (RRRC, http://www.rrrc.us/) maintains a range of research resources including specific pathogen free (SPF) rats, ie, those demonstrated to be free of a defined list of pathogens, as well as live transgenic models, cell lines, and a broad array of cryopreserved rat models (11).

Size and Longevity

One of the major criteria in selecting animal models to study aging is their lifespan because the longer a study species lives, the more expensive and longer to perform longevity experiments become. A common misconception is that because of their larger size, rats are substantially longer-lived than mice. However, this is generally not the case (Table 2). As can be seen, the maximum lifespan of laboratory rat and mouse strains is similar, with some strains of rats even being somewhat shorter-lived than some mouse strains. Note also that the body weight of common laboratory rat strains used in aging research is approximately 10-fold greater than that of mice. Consequently, as body size and longevity are positively correlated among mammalian species, rats are actually shorter-lived than mice once body size is accounted for, although both species are relatively short-lived for mammals. In fact, rats live only 45% as long as an average mammal of the same body size, whereas mice live approximately 70% as long as expected (12). As mentioned earlier, one clear advantage of rats’ larger size is the amount of tissue potentially available, which is important when studying specific tissues, for instance, regions of the brain, as well as for greater ease of conducting surgical manipulations.

Table 2.

Longevity and Weights of Some Laboratory Strains of Rats and Mice Commonly Used in Aging Research (23)

Rat Genotype	Median Lifespan	Maximum Lifespan	Max Body Weight	Mouse Genotype	Median Lifespan	Maximum Lifespan	Max Body Weight
Fischer 344 (F344)				C57BL/6 (B6)
Male	24.0	28.9	460–470	Male	28.0	35.0	40–43
Female	27.1	34.5	340–350	Female	27.1	34.3	30–33
Brown Norway (BN)				DBA/2 (D2)
Male	30.1	36.9	470–480	Male	20.5	30.3	32–34
Female	31.0	38.0	230–240	Female	18.0	30.3	22–24
F344 X BN F1				B6 X D2 F1
Male	33.8	40.6	630–640	Male	32.2	42.0	45–48
Female	32.0	39.2	340–350	Female	29.9	38.0	30–33

Note that these data come from the NIA/NCTR biomarkers study in which all animals of both species were maintained under identical conditions. Lifespan is in months and weight is in grams.

Historical Roles of Rats in Aging Research

Establishment of Rodent Models in Aging Research

The NIA has had an enormous impact on the choice of what strains of rats and mice are used in aging research. At its creation in 1975, it was realized that NIA’s funded research would be facilitated and have its reproducibility enhanced, if the NIA could provide researchers with aged animals raised in clean, standardized conditions. NIA initially provided two rat strains, the outbred Sprague Dawley (SD) and inbred Fischer 344 (F344), plus three mouse genotypes. However, the SD rats were tumor-prone at a relatively early age and susceptible to pneumonia, which compromised their usefulness (13). Use of the SD rat dropped to such a low level that their availability was discontinued by the early 1980s. To a lesser extent, F344 males were overly prone to testicular tumors (interfering with hormonal status on aging) and renal pathology (14). However, because the F344 was by far the most commonly used rat strain, and its aging biology had been so well-documented, the NIA decided to try hybridizing the F344 to five other strains (Buffalo, BN, Hooded, Lewis, and Wistar) to see whether any of the hybrids fulfilled the criteria of avoiding excessive early-life diseases while developing a broad range of pathologies with age as do people. By far, the best of these hybrids was the F344 x BN. Since then, NIA has provided aged animals of three genotypes (F344, BN, and their F₁ hybrid F344BNF1) (13).

Dietary Restriction

For more than half a century, dietary restriction (DR)—reduction in food intake—provided researchers the only method of reliably extending life, an invaluable tool for aging research. Based on his postdoctoral studies on the nutrition of trout Clive McCay hypothesized, correctly as it turned out, that retarding growth might increase longevity (15). In his initial study, the DR began when the pups were weaned, and he restricted the food of these “white rats” such that “the allowance of feed was thus adjusted to hold the body weight of each member of the retarded groups as nearly constant as possible.” In his initial 1935 report on the effect of DR on lifespan, the impact was sex-specific. Mean longevity of males was increased ~70%; however, there was no measurable effect in females. In a subsequent study, he observed a significant lengthening of life in both sexes, something that has generally been reported ever since.

In the 1960s and 1970s, outbred SD rats (mostly male) were used in studies defining how various diet compositions, degrees of DR, and times of restriction initiation affected longevity (16,17). Generally, these studies revealed that the greater the degree and longer the time of restriction were, the greater the life-extending effect it had. A key discovery in this period was that DR reduced tumor burden and deaths from cancer (18,19).

Research from two investigators, Roy Walford and Edward Masoro, during the 1980s was largely responsible for the research community accepting that DR retarded aging in rodents when they showed that DR reduced a variety of age-related pathologies and improved most physiological functions in both rats (Masoro) and mice (Walford) (20).

Masoro’s group published the first report of DR under SPF/barrier conditions, ie, the animals were maintained under HEPA-filtered air (21). These rigorous husbandry conditions accomplished several things: (i) minimized the exposure of animals to selected infectious or potentially infectious agents; (ii) by doing so, enhanced lab-to-lab and study-to-study replicability; (iii) still allowed for an intact microbiome; and (iv) allowed investigators to determine whether a manipulation, such as DR, increased lifespan by retarding fundamental aging processes rather than by increasing an animal’s resistance to infectious agents. In addition, Masoro employed a 40% restriction, ie, the rats were fed 60% of the food consumed by the rats fed ad libitum. This level of DR has become the standard for almost all DR studies over the past three decades.

Although several groups had studied the effect of DR on various strains of mice from 1940 to 1960, the longevity effect of DR on lifespan of mice in these studies was modest (or not significant) compared with the effect DR had on rats (20,22). However, subsequent studies found that usually, although not always, mouse longevity was increased by DR.

Is the magnitude of DR-induced longevity extension similar in rats and mice? This question potentially affects the relative utility of the two species for DR studies. Swindell concluded from a comprehensive review of published studies that DR generally extended the lifespan of rats more than mice (22). It should be noted, however, that his review included a diversity of genotypes, husbandry conditions, diets, and restriction amount. It should also be noted that a number of mouse studies found no impact of DR on longevity.

A more direct comparison of the effect of DR on mice and rats comes from a study conducted by The National Center for Toxicological Research (NCTR) as part of the NIA’s Biomarkers Aging Program. Both sexes of three genotypes of rats (F344, BN, and BNF344F1) and four genotypes of mice (C57BL/6N, DBA2/N, B6D2F1, and B6C3F1) were studied under identical conditions and with identical degrees of DR. In all cases, DR lengthened life. The median increase for rats ranged from 14% (female F344 rats) to 36% (female BNF344F1 rats) compared with a range in mice of 15% (male DBA2/N and C57BL/6N mice) to 52% (DBA2N female mice). The average increase in median survival was 23% for rats compared with 31% for mice (23). Taken together it seems apparent that while DR generally increases survival in both rats and mice, the variability of the response is considerably greater in mice.

Sex Differences in Aging

Human females live longer than human males under virtually all conditions and die at a lower rate from virtually all of the top causes of death (24). The mechanism(s) underlying this robust sex difference in aging remain largely unknown and, in fact, largely unexplored. In mice, sex differences in longevity are highly variable. In a review of 118 mouse studies, Austad reported that in 55% of the studies, males lived longer, and in 43% of the studies, females lived longer (25). This variation was not strain-specific as 29 longevity studies of the most commonly used strain, C57BL/6, had nearly the same range of survival differences as that seen in all strains combined. In rats, there is a much more consistent female survival advantage. It is particularly informative that females uniformly outlived males in the three genotypes used in the NIA Biomarkers study (F344, BN, and F344xBNF1) because all three genotypes were maintained under identical husbandry conditions (26).

In recent years, an astonishing discovery in mice is that drugs or even genetic manipulations that extend life and health often affect only one sex or affect one sex considerably more than the other (27). If this pattern extends to humans, it has major implications for the development and use of health-extending medical interventions. It would seemingly be advantageous to evaluate sex-biases discovered in mice in a second species, especially a second species with sex-biased longevity similar to humans.

Hormonal Effects on Aging

Historically, the initial research on endocrine regulation of aging was conducted in rats. The seminal experiments were conducted by Everitt using a hypophysectomy (HYP) procedure wherein the pituitary was removed in its entirety (28). HYP in young male Wistar rats reduced age-related markers (retardation of aging of tail tendon collagen and prevention of proteinuria) but actually shortened lifespan. However, when HYP was accompanied by lifelong adrenocorticotropic hormone (ACTH) replacement, lifespan was increased (28). In 1977, Joseph Mietes’ group published a paper (29) demonstrating that “starvation” in rats (25% ad libitum feeding) decreased circulating levels of at least five anterior pituitary hormones including luteinizing hormone (LH), follicle-stimulating hormone (FSH), thyroid-stimulating hormone (TSH), prolactin (PRL), and growth hormone (GH). Furthermore, this effect was not the result of an inability of the anterior pituitary to secrete these hormones, but through reduced hypothalamic stimulation. In 1980, Everitt then questioned whether “the anti-aging effects of food restriction may be due to reduced secretion of pituitary aging factors” (30). His group directly compared DR versus HYP + ACTH initiated at either 70 (young) or 400 (middle-aged) days of age in male Wistar rats; both conditions extended lifespan in “young” group. Thus was born the idea that multiple anterior pituitary–secreted hormones might be both beneficial and deleterious to life- and healthspan.

Here, we leave the rats behind for a bit and jump to the nematode, C. elegans, in which a host of genetic discoveries has revolutionized aging research. In 1988, Friedman and Johnson reported that mutation of a single gene they called age-1 was reported to extend worm life by 50% (31). Five years later, Kenyon reported a second worm gene, daf-2, that when mutated extended life even more—two-fold (32). Ruvkun eventually identified age-1 as the worm version phosphoinositide-3-OH kinase (PI3K) (33) and daf-2 as an insulin/IGF receptor (34). So remarkably, both genes were in the same biochemical pathway, homologous to the mammalian insulin–insulin-like growth factor 1 (IGF-1) signaling pathways.

DAF-2 is the lone worm receptor than is homologous to both mammalian insulin and IGF receptors. Upon activation by a ligand, DAF-2 binds an insulin-like substrate to initiate a cascade of events, including activation of the age-1 homolog of mammalian phosphoinositide-3-OH kinase (PI3K). This kinase, in turn, is believed to activate the Akt-encoded protein kinase B (PKB). PKB is a serine–threonine kinase believed to antagonize a forkhead–winged helix transcription factor (daf-16), which is required for extended lifespan in both daf-2 and age-1 mutants. This led to the idea that insulin and/or IGF-1 signaling is critical for increasing lifespan in other organisms.

In mammals, circulating IGF-1 is modulated by growth hormone (GH), which is secreted in pulsatile bursts from the anterior pituitary gland, a pattern that is necessary to achieve full biological activity. GH binds with high affinity to its receptor found in tissues throughout the body, and activation of this receptor stimulates the synthesis and secretion of IGF-1. Although 90% of circulating IGF-1 is synthesized and secreted by the liver, many types of cells, including some found in the brain and vasculature, are capable of local IGF-1 production (29). This knowledge led to the hypothesis that GH is the critical anterior pituitary hormone for the lifespan-increasing effects observed by Everitt in his HYP experiments.

This hypothesis received strong support from two mouse studies. Both the Snell and Ames dwarf mice lacked GH (as well as TSH and prolactin), due to single-gene mutations in developmental transcription factors that prevent the development of the appropriate secretory pituitary cell types. Both dwarf genotypes turned out to be substantially longer-lived than littermate controls (35). Although investigators generally attribute the increased lifespan of Ames and Snell dwarves almost exclusively to reduced GH (to a lesser extent, IGF-1 deficiency), these results can have multiple determinants. For example, hypothryroidism and the reduction in basal glucose levels found in Snell and Ames dwarves lead to a reduction in insulin concentrations that may exert independent effects on longevity (36). Smaller body size alters cardiovascular parameters and energy requirements, and these factors may influence oxidative stress, independent of GH or IGF-1 deficiency. Finally, increased levels of glucocorticoids found in Ames and Snell dwarves further complicate the task of determining whether increases in lifespan are mediated specifically by changes in the GH–IGF-1 axis. Since that time, multiple strains of constitutive and conditional genetically modified mice with reduced GH axis activity have demonstrated increased longevity. However, these animals also appear to have some of the same metabolic issues as described in the Snell and Ames dwarf mice (see ref. 37 for review elsewhere and beyond the scope of the current review).

An even more critical question using these models is whether GH deficiency at specific times of life mediates longevity. Indeed, GH is known to have permanent organizational effects to the central nervous system during critical periods of early development. In mice, 6 weeks of early life (2–8 wk of age) GH injections abolish the longevity phenotype of Ames dwarf mice, hinting that there are critical periods for GH’s life-extending effects (38). To address the various complications in interpretation of GH manipulation in mice and assess the generality of the GH-axis longevity effects, Sonntag and colleagues (39) developed a rat model, the Lewis dwarf, that exhibits a specific and limited GH and IGF-I deficiency in adulthood and assessed effects on pathology, lifespan, and several functional markers of aging. These authors demonstrated that a specific and limited deficiency of GH and IGF-I alone does not increase lifespan but regulates age-related pathology—at least in the Lewis dwarf. In addition, they demonstrated the importance of the peripubertal rise in these hormones for regulation of lifespan and the importance of GH and IGF-I during adulthood for maintenance of tissue function (40). The bottom line is that multiple models are needed so that we may address strengths and weaknesses of individual models to address the question. In fact, constitutive deficiency of GH/IGF throughout life in mice, such as in growth hormone receptor knockout (GHRKO) mice, has salutary effects on aging, but in rats, early GH/IGF life-long deficiency is detrimental and may parallel some issues in GH-deficient humans better than is recapitulated in mice. Such disparate effects between rodents highlight the need for this and other pathways to be investigated more carefully in more than a single mammal species. The rat is ideal for a “second-species” test.

Advantages of the Rat for Aging Research

Similarity to Human Age-Related Pathology

For aging studies, especially in those involving interventions, it is very important to know the pathology that is associated with aging in the choice of animal model. This is especially important in experiments testing the effects of genetic, nutritional, or pharmacological interventions on lifespan and healthspan because the manipulation one is studying could affect one type of age-related pathology, eg, cancer, but not other pathologies that are important in aging, especially in humans.

Laboratory rats and mice have a major advantage over invertebrate models in that they have the same organs and tissues as humans plus there is an extensive literature on histopathological lesions associated with end-of-life, much of it generated from toxicological studies. However, much of these data are of limited value to aging community because the mice or rats studied are not widely used in aging (eg, ref. 41) or the pathological data focus primarily on neoplasms (eg, refs. 42 and 43). In an excellent commentary by Snyder and colleagues (44), discussing the cause-of-death analysis in rodent aging studies, the authors ranked the leading causes of death in four mouse strains and three rat strains. However, they did not report the percentage of animals dying of these various diseases. Therefore, we reviewed the literature focusing only on studies in the past 25 years that: (i) had used relatively large numbers of rats or mice in aging studies and (ii) the colonies of animals were maintained under clean conditions (eg, barrier facilities) as shown by lifespan data either in the paper describing the pathology or from previous papers. This ensures that the pathological lesions described in these studies did not arise from shortening of lifespan by infectious agents or stress.

Table 3 presents end-of-life pathological data from 15 studies; eight for mice (five different strains or stocks) commonly used in aging studies and seven for rats (four different stocks or strains). Probable cause of death can be problematic because of the inconsistency of diagnosis among pathologists and because of euthanizing so-called moribund animals (44). Criteria for euthanasia are somewhat subjective and vary by institution and attending veterinarian, making it often difficult assign a potential cause of death with confidence. Therefore, many pathologists prefer to present pathological data as the prevalence or incidence of the lesions at death. In Table 3, we present the major pathological data as they are reported in the publications, ie, either the incidence or probable cause of death.

Table 3.

Comparison of End-of-Life Pathology in Rats and Mice to Humans

	n	Cancer		Heart Disease		Renal Disease
		Incidence	Death	Incidence	Death	Incidence	Death
Mice
♂ C57BL/6 (47)	58	—	80%	—	—	—	2%
♂ C57BL/6 (85)	45	—	62%	—	—	—	—
♀ C57BL/6 (85)	41	—	73%	—	—	—	—
♂ C57BL/6 (86))	82	—	66%	—	—	—	2%
♂ C57BL/6 (48)	49	83%	57%	—	—	49%	0.04%
♀ C57BL/6 (48)	57	84%	60%	—	—	79%	12%
♂ B10C3F1 (87)	68	87%	—	—	—	—	—
♂ AfC57BK.6F1 (87)	64	89%	—	—	—	17%	—
♂ UM-HET3 (88)	310	—	78%	—	2%	—	—
♀ UM-HET3 (88)	623	—	93%	—	1%	—	—
♂ Ames dwarf+/+ (89)	21	95%	—	—	—	5%	—
Rats
♂ F344 (90)	71	81%	—	62%	—	81%	—
♂ F344 (91)	111	74%	—	58%	—	82%	—
♂ F344 (92)	40	—	38%	—	8%	—	40%
♂ F344 (93)*	20	—	48%	—	3%	—	3%
♂ Sprague-Dawley (93)*	20	—	48%	—	3%	—	3%
♀ Sprague-Dawley (93)*	20	—	57%	—	2%	—	—
♂ Lewis (84)	47	88%	57%	—	2%	—	74%
♂ Brown Norway (94)	90	NG	—	63%	—	35%	—
♂ BNF344F1 (94)	220	NG	—	69%	—	32%	—
♀ BNF344F1 (94)	150	NG	—	9%	—	14%
Humans
United States†		38%	21%	48%	24%	15%	2%

The percent of animals (males or females) showing either the presence (incidence) of specific pathology or the likely cause of death (death) at the end of life is given for the various strains of mice and rats where the data were reported. The variation in values represent differences in male and female animals.

Abbreviations: n: number of animals on which pathology was conducted; NG: Data were not given for number of rats with neoplasia; rather, the pathology was presented for specific tissues.

*Data from spontaneous death up to 26.5 months of age.

^†Current data are presented for men and women combined and were obtained from the National Cancer Institute (cancer) (https://www.cancer.gov/about-cancer/understanding/statistics), Centers for Disease Control and Prevention (heart disease) (https://www.cdc.gov/heartdisease/facts.htm), and National Kidney Foundation (renal disease) (https://www.kidney.org/news/newsroom/factsheets/KidneyDiseaseBasics).

As Table 3 shows, the cancer is the major end of life pathology found for all mice and rat genotypes. In mice, cancer incidence ranges from 83% to 95% and in rats from 74% to 88%. Note that this incidence is not indicative of what animals die from—only what they die with. In inbred rodent strains, one type of cancer often predominates. For example, testicular (Leydig cell) tumors are observed in almost all old F344 rats but is found in less than 5% in Long-Evans, Sprague-Dawley, and Wistar rats (45), and large granular lymphocyte leukemia is a major cause of death for F344 rats, affecting over 50% of the males (46). In contrast, lymphoma is the primary neoplastic lesion usually resulting in the death of over 70% of C57BL/6 mice (47,48).

The UM-HET3 mouse stock used in the NIA-funded Intervention Testing Program is a defined colony of genetically heterozygous mice derived from F₂ mating of four inbred lines (BALB/cj, C57BL/6J, C3h/HeJ, and DBA/2J). These mice show the occurrence of nine different types of tumors. However, the probable cause of death in ~80% of these male and female mice is also cancer (Table 3). Therefore, the incidence of cancer is a major pathological lesion in laboratory mice whether it occurs predominantly in one type of tumor as in inbred lines or in multiple types of tumors as in UM-HET3 mice. In comparison to humans, the incidence of cancer at end of life is roughly twice as high in laboratory rodents compared with humans (Table 3).

Although cancer incidence is similar in laboratory rats and in all mouse genotypes, the appearance of major non-neoplastic lesions is quite different for rats and mice. In mice, ~20% of the deaths that arise from non-neoplastic lesions are diverse with no lesion making a major contribution to the pathological status of the mouse. In contrast, heart and renal pathology are major pathologies in rats (Table 3). For example, only one study with UM-HET3 mice showed any evidence of chronic cardiac pathology. In contrast, all of the strains of rats given in Table 3, except Lewis rats, showed a high incidence (35% to 69%) of chronic cardiomyopathy. Interestingly, this incidence of heart disease is similar to that found in humans. Renal disease is also a major disease in all four strains of rats listed in Table 3. Renal disease was also observed in mice but only in five of the eight the mouse studies reported in Table 3. The incidence of renal pathology in rats much greater than found in humans.

In summary, some laboratory rat genotypes and all laboratory mice exhibit a high frequency of tumors at the end of life; however, there is a major difference in the occurrence of major, nonneoplastic lesions, which is also observed in humans. Thus, one would predict that manipulations that affect aging might act differently in rats and mice depending on the pathology that the manipulations target.

Insulin Resistance

In humans, insulin resistance is considered an important component of metabolic syndrome and a significant risk factor for other metabolic and tissue defects, including Type 2 diabetes (49). Therefore, preserving or restoring insulin sensitivity is an important target of lifestyle modifications and therapies to delay aging and age-related diseases (50,51). Likewise, insulin action in rodents is often considered an important readout of healthspan in normal aging and disease models (52). Although caveats regarding the role of insulin resistance in longevity determination exist in mice and lower organisms (53), insulin sensitivity is a hallmark trait of many long-lived dwarf and dietary-restricted models (52). Euglycemic clamps are considered the “gold standard” method for assessing insulin action in rodents and humans (54). In mice, clamp studies are elegantly and routinely conducted by the skilled personnel in core facilities and a handful of laboratories to provide valuable insights into our understanding of metabolism (55,56). Indeed, mice have been the preferred model for such studies due to their heavy use in basic research as well as the unique advantage that transgenic mice have provided to address mechanisms (57,58).

Although mice offer notable value for metabolic studies, there are important drawbacks and disadvantages to the mouse as a model for studies assessing glucose metabolism, compared with rats. Physiologically, a typical 3 mU•kg⁻¹•min⁻¹ hyperinsulinemic clamp in young healthy rats will concurrently suppress hepatic glucose production (HGP) by a modest ~50% and increase glucose uptake (Rd) by approximately twofold across strains, thereby allowing for simultaneous assessment on glucose production and disposal in the same clamp study (59–65). However, the optimal insulin dose in mice can often vary by strain and sex, and range from ~1 to 4 mU•kg⁻¹•min⁻¹ (57). As such, mouse clamps often necessitate optimizing the insulin dose that best suits the needs of the investigator, and identifying a dose that simultaneously provides useful data on glucose production and uptake can be challenging. For example, using 1–1.5 mU•kg⁻¹•min⁻¹ doses may lead to ~50%–75% suppression in HGP in control mice, thereby allowing for the effects on liver to be evaluated, but fail to stimulate glucose uptake. On the other hand, a 4 mU•kg⁻¹•min⁻¹ clamp can robustly stimulate Rd by two to threefold in mice, but occurs at the expense of near or complete suppression in HGP (55). It is also worth noting that when comparing basal HGP and maximal stimulated Rd across species, stark differences exist between mice and humans, whereas rats sit somewhat closer to humans than do mice (Table 3).

Technically, mice also pose unique challenges over rats, such as placement of the vascular catheter for the surgeon, as well as the ability to maintain patency of the lines during surgical recovery. Additionally, without previous experience in the mouse model to be studied, the technician may find it initially difficult to gauge how quickly and aggressively to “ramp up” the glucose infusion rate upon insulin administration, and mice tend to be less forgiving to instability in glucose levels from inadequate adjustments in the infusion rate during a clamp than rats. Importantly, due to their larger body size, rats pose fewer issues related to the infusion of tracers and other compounds and the frequency and volume of blood sampling, and upon terminal collection of blood and tissues, rats afford the opportunity to measure a greater number of endpoints from the same animal. Thus, given the increasing focus of geroscientists to evaluate the metabolic impact of geroprotective compounds and strategies in normal animals, and the increasing availability and feasibility of generating transgenic rat models, investigators should strongly consider incorporating the rat model for sophisticated in vivo assessment of glucose metabolism (Table 4).

Table 4.

Approximate Basal Glucose Production and Max Stimulated Glucose Disposal Rates in Human, Rat, and Mouse Clamps Studies

	HGP (mU•kg−1•min−1)	Max Rd (mU•kg−1•min−1)
Humans	~2	~10–12
Rats	~8–12	~20–25
Mice	~15–20	~60

Behavior as Elements of “Healthspan” or Why Use the Water Maze to Test Memory?

In the last 20 years or so, there has been a growing acceptance of the relative importance of considering healthspan (the period of life free of disease and optimal behavioral function) rather than only lifespan in the context of intervention studies. In 2014, a paper (66) published in Cell stated:

Basic aging research has always been intriguing, but until recently seemed to hold few concrete solutions to advance health. Despite limited resources, the field recently erupted and now offers routes to achieve a new and ambitious goal–a substantial extension in health life expectancy, or “healthspan”.

This has led to the emergence of geroscience as the driving approach in the basic biology of aging. Included in this definition are fundamental aspects of behavioral functioning including cognitive outcomes that certainly decline during the aging process. Studies incorporating such functional measures are on the rise but suffer from a lack of rigor and knowledge of the neuroscience of aging. In fact, mice are not “little rats” and differ along many domains of function that are central to assessing healthspan measures. Thus, healthspan outcomes, as they are measured in mice, may tap into other domains of mouse ethology and biology because of differences in social behavior, neuroanatomy, and even the nature of neurodegenerative diseases being studied. While a complete review of all the behavioral differences between mice and rats is not possible in this review, we highlight those that have been specific to the field of the biology of aging and which may have implications for translation to humans (also see ref. 67 for a great review).

Social behavior

In terms of social behavior, although both rats and mice live in large groups, rats are much less territorial. In fact, rat males mate with all females (68,69). On the other hand, mice live in territorial structures (called “demes”) with a single male mating with multiple females (70). Mice burrows are thus are less complex than rat burrows, with a single cavity occupied by a one male (71). Thus, confrontations between male mice are rarer and are more aggressive and territorial. Therefore, the traditional housing of multiple male mice in any lifespan, healthspan, or intervention study may be influenced by social stress and may have an enormous impact on measures of cognitive function.

Cognition

Cognitive deficits, including those in learning, memory, attention, and executive function, are a core feature of many neurological disorders especially in the context of aging and in the context of pathological aging, including neurodegenerative diseases such as Alzheimer’s disease. The ability to develop preclinical models and testing methodologies of these conditions is critical to the ability to translationally develop interventions to circumvent these conditions. There is a long history in the sciences of psychology and neuroscience that established such approaches are steeped in using rats as subjects and are not necessarily known to investigators studying the basic biology of aging.

The rat was used in learning and memory experiments by Willard Stanton Small, the first developer of the “maze,” published in 1901 in the American Journal of Psychology, which was founded in 1887 by the famous experimental psychologist Stanley Hall. The paper was titled “An experimental study of the mental processes of the rat.” In this seminal paper, Small describes using the ethological ability of the rat to navigate a complex maze to receive a food reward.

The chief difficulty in such experimentation lies in controlling the conditions of the problem without interfering with the natural instincts and proclivities of the animal, and thus distracting or deflecting its attention. An animal should be made to do difficult things only in line with its inherent abilities (72).

In this context, Small took advantage of the rat’s “natural” burrowing behavior to explore the maze and extrapolated studying memory in rodents to help explain the process of memory in humans. However, Small relied on food restricting animals as motivation to run in mazes, a situation that confounds studying aging, given that DR itself affects aging.

A work-around this issue was the development of the Morris water maze, described by Richard Morris in 1981 (73). This assay has become the gold standard in aging research to investigate learning and memory. Animals are placed in a circular pool filled with water in which a platform is submerged just below the surface of the water. The water is made opaque and the platform hidden from view by placing white dye into the water. By placing the animals in different starting positions within the pool, they are trained to find the hidden platform based on the assumption they make use of external cues to circumnavigate to the escape platform. Here, the motivation in this assay is the desire to escape the water rather than to obtain a food reward, thus making DR unnecessary.

Compared with rats, mice perform more poorly in this task perhaps based on species differences including the fact that rats burrow near water and spend time swimming while mice do not; however, in the same study, mice did just as well in a land-based version (74). As suggested earlier, other noncognitive contributors, such as increased anxiety, may have a larger influence in mice. For example, Lipp and Wolfer (75) that learning performance in mice is related to a preferred strategy of thigmotaxis (swimming around the outside walls of the pool) as opposed to using a spatial learning strategy that is dependent upon brain areas impacted by aging (eg, the hippocampus and prefrontal cortex).

Thus, rats are a preferred species for investigating age-related declines in learning and memory. Their performance also appears to be less adulterated by noncognitive contributors than are mice and the tasks that have been prioritized for assessing cognition during aging such as the water maze were designed specifically with the rats’ natural instincts and proclivities (72) in mind.

Models for Neurodegenerative Diseases: Example of Alzheimer’s Disease

Alzheimer’s disease (AD) is the most common cause of dementia leading to a devastating impact on memory and is characterized by brain pathology including plaques and tangles (76). A growing literature has also investigated the contribution of vascular causes although the differences between mice and rats have been less rigorously tested and thus beyond the scope of the current review.

The identification of mutations in genes such as APP, PS1, and PS2 that invariably led to early-onset AD supported the “amyloid hypothesis” of the pathological cause of AD because all of these genes are involved in amyloid production and processing. According to the ALZFORUM (alzforum.org), at least 172 AD mouse models have been developed on the basis of these genetic associations leading to over 300 human trials of drugs with some therapeutic effect in mice. All of these trials have failed, causing some researchers to question whether mice are informative animal models for this disease (77). In the last 10 years, however, at least eight rat models have also been developed. This is transformational because, in addition to the aforementioned benefits of rats over mice in cognitive tests, rats, like humans, have six isoforms of the tau protein, whereas mice have only four (78). Hyperphosphorylation of tau is involved in the formation of tangles, an essential pathological hallmark of AD, and the similarity in isoforms between humans and rats could indicate a higher degree of similarity in tangle formation as well, although rats do not develop plaques. For example, the Tg6590 rat model does not present with mature plaques, (79) whereas the Tg2576 mouse (similar mutation) does (80). However, cognitive deficits are present in both rat and mouse models, which together with the failure of clinical trials of drugs that reduce amyloid plaque formation, calls into question the amyloid hypothesis. In addition, the more recent TgF344-AD rat model (available from the RRRC) provides a seemingly improved model of brain amyloidosis, in that these rats manifest an impressive spectrum of age-dependent AD-related pathologies including amyloid plaques surrounded by activated astrocytes and microglia, neurofibrillary tangles with hyperphosphorylated tau, and importantly neuron loss. No mouse model combines this array of true AD features (81).

In summary, the development and ultimate strength of any scientific set of hypotheses must rely on converging evidence from multiple sources including other strains and species, including genetically altered models. However, in terms of behavior and physiology, the rat is an excellent choice for evaluating changes in cognitive function and neurodegeneration. Indeed, an incredible amount of time and resources have been invested over the last 40 years in search of a cure for AD using mouse models, and all of which have failed to translate to humans (see ref. 82 for an excellent and extensive review). Thus, the rat should be considered as a potentially highly information additional animal model for the mechanistic understanding and development of therapeutic approaches for AD.

The Path Forward

We feel it is time to bring the rat back to the forefront of aging research. We have enumerated many but certainly not all of the advantages of rat models. We are not suggesting that they replace mice but serve as an informative complement to them.

The one big advantage that mice had over rats—a genome that could be precisely manipulated—has largely been eliminated by the development of CRISPR/Cas technology. In addition to the clear advantages of the rat, it is scientifically useful to have a second mammalian species to better evaluate the generality of mouse findings. For instance, it is at the very least provocative that reports on the impact of reduced GH signaling on longevity in rats appears to contradict a wealth of findings in mice (83,84). It would also be worth comparing the results of life-extending drugs discovered in mice by the Interventions Testing Program in rats as well because sex differences and pathology in rats are more similar to humans than mice. Similar results in two species provide a much stronger rationale for moving an intervention to human studies than results from a single mammalian species, as shown by the rather poor translatability of findings in mice to humans in cancer interventions and the spectacularly poor translatability with AD interventions. In fact, the tremendous advantage of rats for studies relying on cognitive metrics makes one wonder whether rats ought to be the mammalian model of first choice in age-related cognitive decline and AD.

So as we celebrate the Year of the Rat in 2020, at a time when the global population is aging more rapidly than at any time in human history, it would be fitting to give serious consideration to returning the rat to its rightful place in aging research.

Funding

This work was funded by National Institute on Aging (Grant/Award Number: P30AG050886 to SA, P30AG050911 to AR, and R01AG054538 to CSC).