Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Heterogeneous Stocks and Selective Breeding in Aging Research Gerald E. McClearn and Roger J. McCarter Abstract There are advantages and limitations to using genetically heterogeneous stocks and selective breeding procedures in gerontological and healthspan research. Animal models that address complex systems of aging involve constraint or manipulation of genetic diversity. They derive from levels of genetic analysis ranging from molecular to quantitative and are relevant to levels of causal hierarchy from base sequences to complex multivariate phenotypes. For some research purposes control of the genotypic source of phenotypic variability by fixation is desirable; others involve establishing genetic diversity or deliberately manipulating identified genes or anonymous gene complexes to specification. Genetic heterogeneity is essential or advantageous for multivariate description of complex phenomena, the examination of associations among variables, or manipulation of polygenic systems. This review concentrates on these latter quantitative requirements. Space limitations preclude a comprehensive review, but relevant sample references and some hints of historical perspectives are provided, with apologies to the many relevant authors not cited. Key Words: aging; animal model; complexity; genetic variability; genotype; heritability; heterogeneous stock; phenotypic selection; selective breeding Genetic Perspectives on Health and Aging: Genes as Variables K nowledge of the nature of aging owes much to a very substantial body of data derived from the study of animal model systems. Because of convenience, economy, and the wealth of available biological information about them, there has been a concentration of research on rats and mice. Although a strong case can be made for the importance of broadening the phyletic base of gerontological knowledge (Austad 1993), it is likely that these species will remain an Gerald E. McClearn, PhD, is Evan Pugh Professor of Health and Human Development and Biobehavioral Health, and Roger J. McCarter, PhD, is Professor of Biobehavioral Health, both in the College of Health and Human Development at the Pennsylvania State University in University Park. Address correspondence and reprint requests to Dr. Gerald E. McClearn, College of Health and Human Development, The Pennsylvania State University, 315 Health and Human Development East, University Park, PA 16802 or email [email protected]. 16 important source of knowledge for the foreseeable future. Broadly described, these models variously involve stable, homogeneous genotypes, genetically heterogeneous stocks, and directionally manipulated genotypes. In the context of experimental design, the genes may serve as controlled variables, randomized variables, or independent, manipulated variables. Inbred Strains Perhaps the most extensively used control procedure has been the use of inbred strains. (We provide a brief summary as a basis for comparative evaluation of models generated by various mating regimens; for an in-depth discussion, we refer the reader to Yuan et al. 2011 in this issue.) Inbreeding involves the mating of relatives—in laboratory animals, usually siblings. The effect of mating a female with a brother is to increase the genetic similarity of the offspring relative to that of animals derived from random mating. Over successive generations, the level of homozygosity in likeallelic configuration at all loci asymptotically approaches 100% (Falconer and Mackay 1996). Another strain similarly derived will also be “isogenic” but for a different set of alleles. With respect to genotype, inbreeding is not deliberately directional. It is important to stress, however, the natural selection of inbreeding depression: reproductive fitness generally decreases as homozygosity increases, and a substantial proportion of nascent inbred strains terminate through early death, infertility, or impaired rearing, among other reasons. The roster of standard inbred strains must be regarded as select survivors of the inbreeding process and can hardly be considered representative of the species as a whole. Furthermore, the attainment of near-complete homozygosity is no assurance of reduced phenotypic variance. Indeed, an active research area explores the increase in sensitivity of inbred strains to environmental effects due to their elevated level of homozygosity (see Petkov et al. 2005 for a description of constraints on distributions of alleles in functionally related genes). In quantitative genetic analysis, individual differences among animals of a strain are attributed to environmental sources, which may be shared or nonshared among the subjects (despite rigorous experimental controls, animals can be differentially exposed to environmental subtleties: e.g., number of siblings in the cage, intrauterine implantation site, height of cage rack shelf, time of day of phenotype measurement, exposure to olfactory stimuli, handling trauma). Moreover, as discussed below, the effects of a particular environmental ILAR Journal influence may vary depending on the genotype of the individual. Measurement error is also usually attributed to environmental interindividual variance. The inbreeding process in its simplest application is unconcerned with any specific phenotype (except fitnessassociated traits); it concerns relatedness of mates only. Thus, after the requisite number of consecutive generations of inbreeding, the genotype approaches uniformity, but of an unknown configuration (males differ from females in having a Y chromosome, of course, but, like other chromosomes, it will be genetically the same in all males in the strain). The power of inbreeding is remarkable: it generates a (relatively) fixed configuration of a fundamental source of organismal individuality—the genotype—without the requirement of identifying, characterizing, or measuring any element of that genotype. The resulting uniformity has made possible a high degree of standardization with respect to genetics, both across laboratories and within laboratories across time. Of course, this model of stability is subject to effects of mutation. Falconer and Mackay (1996) and Roberts (1967) provide details of quantitative genetics in general and of the process of inbreeding specifically; Silver (1995) presents nomenclature and descriptions of major contemporary strains as well as the history of strain origins. Strains have for decades played a leading role in basic genetic research as well as in extensive biomedical and agricultural applications. The undeniable advantages of standardized genetic control offered by an inbred strain are accompanied by a restricted range of generalization. Yet conclusions from a single inbred strain are sometimes uncritically interpreted as characterizing all members of its species. As far as genetically influenced variation and covariation are concerned, however, the situation is comparable to that of generalizing to a human population from a sample of one. Indeed, Roberts (1967, 248) noted that “an inbred strain can be considered as a representative of one gamete only from the base population.… Any conclusions from such work should not be deemed to apply to the species at large without supplementary evidence.” Multiple Strains One tactic to obtain such supplementary evidence is the examination of other inbred strains. A consistent result across a panel of different inbred strains will build confidence in the generality of a particular outcome; an inconsistent one will prevent premature sweeping conclusions. Such a research strategy capitalizes both on the genetic uniformity within strains and on the genotypic differences among them, even if there is no specific evidence about the number, chromosomal location, or molecular identity of any of the involved genes. Multiple strain comparisons offer vivid examples of the potentially powerful influence noted above of the strain “background” or “residual” genotype on the effect of any particular gene or environmental factor. An exemplar case of influence of such background on the effects of a single “disease” gene is that of Leiter and colleagues (1981), who examined the Volume 52, Number 1 2011 diabetic symptom pattern displayed by mice homozygous for a mutation that occurred in a particular inbred strain. The authors established the mutant allele (recessive, identified as db) in homozygous condition in seven other inbred strains and compared these animals to “normal” animals (homozygotes Db/Db and heterozygotes Db/db) within and across strains with respect to body weight, plasma insulin, blood sugar level, and mortality. The results were dramatic and classical: there were huge differences between strains in the overall values of control animals for all of the phenotypes, and the difference within strains between the “normal” controls and the homozygous recessive animals was enormous. Thus a single “disease” gene may or may not be accompanied by the symptoms of the disease depending on the genetic context. Likewise, an environmental influence may be a function of the genotype of the recipient organism. In the context of research suggesting aluminum exposure to be a risk factor in the etiology of Alzheimer’s disease, Fosmire and colleagues (1993) examined the effects of a high-aluminum diet in five inbred mouse strains. Relative to matched strain mates on control diets, the high-aluminum diet resulted in an elevated brain level of aluminum in only one strain. If the research question concerned whether dietary aluminum can affect brain levels and only this particular strain had been selected for the research, the correct answer would ensue. If the question was whether dietary aluminum does (always) affect brain level, the same choice of strain would have given an incorrect conclusion. A disparity of results as in this case not only recommends modesty in claiming broad generality of results but also can indicate promising avenues for further exploration; for example, the availability of both susceptible and nonsusceptible strains invites study of blood-brain barriers or other mechanisms that might affect susceptibility. Another striking result concerns the influence of caloric restriction on longevity. The life-extending effect of this treatment has long been regarded as highly reliable and dependable. In an examination of 41 recombinant inbred strains of mice (more on such strains below), Liao and colleagues (2010) observed that in some strains caloric restriction produced life extension, in about the same number the influence was life-shortening, and in others there was no appreciable effect. Recently, of course, molecular genetic methods have made possible the detailed identification of specific loci or of the whole genome of inbred strains as well as procedures such as the introduction of transgenes and targeted mutagenesis (Yuan et al. 2011). Thus, for example, the comparison of standard control animals of an inbred strain to those of the same strain but with a single locus manipulation is likely to become more popular, particularly in reductionist exploration of the paths from genes to phenomena higher in the causal hierarchy (“reverse” genetics). This merger of classical and molecular genetics is full of promise. The exquisite control provided by these molecular methods does not, however, eliminate the restraint on generalization of single-strain results. Any finding on a single inbred background simply establishes that outcome as a possible one. It is not at all informative 17 about the generality of the result, but may reveal new research opportunities. F1 Hybrids Matings between different strains produce F1 (first “filial” generation) hybrid offspring, which have a greater allelic diversity than that of their inbred parents by virtue of being heterozygous for each autosomal locus for which the parent strains differ. F1s are also uniform with respect to their autosomal genes, but the males can differ with respect to their Y chromosome, depending on the strain of their male parent. (They may also display effects due to maternal mitochondrial genes and to other “parent of origin” characteristics identified in recent epigenetics research.) A frequent research design involving F1 animals compares them to their parent strains. This design permits evaluation of effects of dominant inheritance, including the phenomena of inbreeding depression and of the “hybrid vigor” associated with differences in heterozygosity. Phelan and Austad (1994) have observed that F1 hybrids are often more biologically uniform than their inbred parents. Genetic Heterogeneity For much research using fixed genotypes, the outcome is describable in terms of group mean differences, where the within-group variance is the source of the error term. As noted earlier, this variance is of environmental (or error) origin only and is thus limited in any investigation in which phenotypic variation or covariation is of central interest. For these latter purposes, groups that segregate at many genetic loci, display the effects of diverse genes, and allow both intralocus effects of dominance and interlocus interaction effects offer many strong advantages. Genetically heterogeneous stocks (HS1) that are systematically derived and maintained provide many opportunities in agricultural and biomedical research—for use as normative groups, for parent-offspring heritability analyses, for quantitative trait locus (QTL1) analyses, for any analyses involving hypotheses about correlations among phenotypes or the generation of multivariate constructs, and for phenotypic or genotypic selective breeding (as foundation stocks). Although the merits of such heterogeneous stocks have long been recognized in many research domains, they have been underexploited in healthspan and gerontological research. A number of papers (McClearn and Hofer 1999; McClearn and Meredith 1964; Miller et al. 1999b; Phelan and Austad 1994) have restated the arguments (including some dissent; Festing 1999) for using appropriate genetically heterogeneous stocks in the study of aging. Because of a growing appreciation of the “complex system” attributes of developmental processes, and the consequent necessity of multivariate analytic procedures, these considerations will certainly make the use 1Abbreviations used in this article: HS, heterogeneous stock(s); QTL, quantitative trait locus 18 of heterogeneous stocks increasingly frequent in gerontological research. There are some genetically heterogeneous stocks for which the founding “recipe” is not known but that may have many of the advantages of heterogeneity just described. Many have become standard features of research in different research domains and have yielded valuable amounts of reference and normative data. Chia and colleagues (2005) provide a detailed and critical review of the origins and uses of such mouse outbred stocks. However, it seems likely that the possibility of reconstituting the foundation stock from intermating of known standard inbred strains will be of special merit as new tools are developed and exploited. F2 Crosses Although all animals in an F1 have near-identical genotypes, they do not afford stability across generations. In the mating of a female and male from the same F1, each may provide the allele from either parent. Thus, there are three possible genotypes in the offspring of F1 × F1 matings for each locus for which the parent inbred strains had a different allele. Each member of the resulting offspring, the F2, will be genetically unique. If there are recurring needs for research animals of the F1 genetic configuration, they must be generated anew by appropriate matings of the inbred strains. Thus F2s present neither genetic uniformity nor constancy across generations. However, relative to inbred strains, they offer enhanced genetic diversity within a generation and can provide relative constancy across generations, not of individual genotypes but of the group’s allelic frequencies. For many purposes, this variability is no shortcoming but a strong virtue: the allelic diversity in a group of F2 animals makes them a better platform for generalization of research results than any single group of homogeneous genotype. Recombinant Inbred Strains Some of the merits of F2s and of inbred strains have recently been merged in the generation of recombinant inbred (RI) strains, which are produced by inbreeding a number of lines from a common genetically heterogeneous parent population. For example, numerous RI strains have been inbred from an F2 population derived from the C57BL/6 and DBA/2 inbred strains, whose allelic diversity is reshuffled into different configurations in the RI offspring. RI strains differ from each other within the limits provided by the initial gene pool diversity, but provide for relative genetic uniformity within strain and stability across laboratories and across time. Thus each strain has the same advantage of genetic uniformity and the same limitation of genetic uniqueness as all inbred strains, but collectively they present an array of standardized unique configurations of the genetic diversity of the two initial parent strains. RI strains have recently featured in the identification of QTLs. The molecular mapping of the mouse genome has made possible the identification of chromosomal regions that are ILAR Journal associated with variability in a phenotype of interest. These regions vary in size and may contain one or more as yet unidentified genes affecting the trait. The nomination of QTLs is often an early step in exploring the causal chain back to the molecular level of the gene. In a landmark application to aging, Vieira and colleagues (2000) explored the association of QTLs to longevity in Drosophila under differing environmental conditions of temperature or feeding. The result was an archetypical demonstration of gene-environment interaction, with many effective QTLs pertinent to only one environment and not others, in one sex but not the other, in different directions in the sexes, or in both sexes but in only one environment. In mice, Gelman and associates (1988) identified six chromosomal regions associated with age at death in a sample of RI strains derived from C7BL/6 and DBA/2 inbred strains, and a recent project examined animals of another sampling of this RI panel as well as F2s from the same gene pool in a longitudinal study of diverse age-related variables (Lang et al. 2010). In other studies researchers have identified QTLs for blood pressure (Blizard et al. 2009), body length, weight, adipose mass, serum alkaline phosphatase, and structural and material properties of the femur and tibia (Lang et al. 2005); and the weight of soleus, gastrocnemius, tibialis anterior, and extensor digitorum longus muscles (Lionikas et al. 2006). The search for QTLs from RIs derived from an F2 can reveal the chromosome regions containing effective genes only if the original inbred parent strains of the F2 possessed different alleles for those genes. Clearly, restrictions pertain to generalization from single panels, so several panels of F2-derived RI lines are in use, with different F2 origins, making empirical confirmation feasible. Furthermore, several projects have established the utility of seeking QTLs from the more heterogeneous gene pools of 4-way-cross mice (Miller et al. 1998; Turri et al. 2001) and rats (Fernandez-Teruel et al. 2002) as well as in other heterogeneous stock mice (Valdar et al. 2003). Increases in the use of RI strains for addressing complex genetic systems are promised by the Complex Trait Consortium (Churchill et al. 2004; Threadgill et al. 2011). The general availability of these strains for biomedical research should have a major influence in the near future on gerontological science (Chesler et al. 2008). Heterogeneous Stocks The demonstrated broad utility of F2 animals suggests that offspring with four rather than two grandparental strains would be even more satisfactory for many purposes. Recently, such 4-way-cross animals have seen increasing application in gerontological research (e.g., Miller et al. 1998, 2002; Swindell et al. 2008; female parents of these 4-way animals were F1 offspring of a cross between BALB/c and C57BL/6 mice, and the males had C3H and DBA/2 parents). The versatility of the approach is illustrated by the fact that these authors addressed relationships between longevity and causal agency at the molecular genetic level (marker loci) and between Volume 52, Number 1 2011 longevity and biomarkers in putative causal pathways of gene influence (body weight, and IGF-1, serum leptin, and thyroid levels). In a further example, Harrison and colleagues (2009) used this 4-way-cross group for evaluating the life-extending effects of rapamycin. Greater heterogeneity is provided by eight-way-cross groups, which have been extensively used in agricultural research. Two such stocks have been useful in studies of aging, health, and cognitive functioning. The (Boulder) HS (McClearn et al. 1970) has been used in longitudinal quantitative genetic analyses and multivariate analyses of a variety of biomarkers of aging from domains of behavior, immunology, physiology, and pathology (Heller et al. 1998; Strauss et al. 1992), and Galsworthy and colleagues (2002, 2005) used this stock to explore psychometric and multivariate aspects of cognitive functioning. The (Northport) HS (Talbot et al. 2003) were experimental models (Valdar et al. 2006a) in projects on fine mapping of QTLs pertinent to human disease and to a variety of immunological, biochemical, and hematological phenotypes. Huang and colleagues (2009) used the same HS to investigate transcript abundance, illustrating the potential for combining quantitative and molecular genetics. The merit of heterogeneous stocks is clearly a function of the number of foundation strains. Also relevant is the number of generations since founding. Initially, there is advantage in an increasing number because the problem of spurious association due to linkage declines as the number of crossing-over events increases, providing enhanced precision in chromosomal location. Later, due to the pragmatically necessary limits of population size, the cumulative effects of inbreeding will have reduced allelic diversity, but this may be offset by the accumulation of normative data over generations. Phenotypic Selective Breeding A particularly valuable use of genetically heterogeneous stocks is as foundation stock for selective breeding programs. Methodologically speaking, much of quantitative genetics depends on associative research procedures in which strong theory permits the interpretation of correlations among relatives in terms of statistical behavior of hypothesized but anonymous groups of genes. Molecular genetic techniques permit the introduction, alteration, or elimination of the influence of specific identified genetic loci. Selective breeding provides the researcher with the capability of directional manipulation of large genotypic complexes, even though the individual loci in the complexes may be entirely unknown. Early Selective Breeding Early humans used pragmatic selective breeding for food species with stupendous success. In terms of scientific research, selective breeding was critical for testing Darwin’s theory of natural selection (Falconer 1992). For more than a century it has been a major method in the application of 19 quantitative genetic theory in agriculture, behavioral genetics, pharmacogenetics, biomedical genetics and many other domains, where it has generated research models of enormous power (Falconer and Mackay 1996; for an overview of principles and applications of selective breeding to pharmacogenetic research, McClearn et al. 1981). Harrison and Roderick (1997) described opportunities awaiting selective breeding for aging research in mice; we address some of these areas in the next section. Selective breeding assigns mates on the basis of similar values of a particular phenotype. In the simplest case, the objective may be to increase the level in the population of some desirable trait. At a rate dependent on the heritability and the extent to which the selected animals differ from their population average (the “selection differential”), the mean level of the phenotype will change over repeated generations. Many research purposes are best served by bidirectional selection, creating high and low lines for comparison with each other and with controls. The resulting model systems are comprehensive, in the sense that all of the gene loci, structural or regulatory, segregating in the foundation population and capable of influencing the phenotype (within the limitations of sampling and inbreeding due to finite sample size) are distributed in alternative allelic form in the upward and downward selected lines. The phenotypic difference between the lines is due to differences in all of the many systems through which the genes operate and for which gene-influenced differences exist in the foundation population. Selectively bred lines are thus powerful research tools for testing hypotheses about the mechanistic routes from the level of the genes upward in the causal hierarchy through the proteome and metabolome to the gerontological phenotype of interest.2 As just one example of the utility range of such a tool, Williams and colleagues (2004) identified some 400 studies exploring the biochemical, biophysical, pharmacological, and anatomical attributes of two lines of mice selectively bred (from the Boulder HS) for high and low sensitivity to hypnotic doses of alcohol (McClearn and Kakihana 1981) and from their derivative inbred lines. Uses of Selective Breeding in Aging Research An exemplar of the power of selective breeding to produce model systems pertinent to aging and health is that of the panel of senescence-accelerated mouse (SAM) strains (Takeda 2009): senescence-acceleration prone (SAMP) and senescenceacceleration resistant (SAMR) strains, developed by selection and inbreeding, have contributed substantially to understanding of early-onset aging processes. An unorthodox feature of these strains is that the foundation stock had been designated “inbred.” The individual differences in early senescence in this strain were so pronounced that the selection regime was initiated, with dramatic results. Clearly, the foundation strain 2Indeed, the potency of selective breeding is captured by the synonyms “experimental evolution” and “laboratory evolution” offered by Rose and colleagues (2006; Burke and Rose 2009). 20 was genetically heterogeneous from an unknown source and to an unknown extent. The merit of selected lines is not restricted to analyses of associated traits. For example, they also constitute valuable test vehicles for assessing the efficacy of proposed interventions to elevate or reduce the level of the primary selected phenotype or any of its covariates. A particularly engaging form of phenotypic selective breeding that addresses issues of environmental influence is that of genetic assimilation. An environmental manipulation is applied to a genetically heterogeneous population, and individuals with resulting like-extreme phenotypes are mated. Then the intensity of the manipulation is reduced and the procedure repeated. The outcome is the derivation of two lines, one with heightened and one with lessened sensitivity to the environmental influence. In the classical genetic assimilation studies on Drosophila by Waddington (1956), the sensitive line eventually required no environmental manipulation for the phenotype to be present. Genetic assimilation has been closely examined in the context of evolutionary theory (Kaneko 2009; Minelli and Fusco 2010) and may have significant potential for studies of environmental risk factors or therapeutic or preventive interventions in aging processes. Another variant of phenotypic selection exerts selection pressure in the usual way but with the effect of a hypothesized component of the mechanistic network controlled. Thus, mates may be selected to produce high and low offspring with respect to an age-related biomarker but with the requirement that there be no mean difference in the two lines with respect to (for example) oxidative stress. Failure to achieve separation between the lines in the primary phenotype would suggest a crucial (necessary?) mechanistic role for the controlled variable. Success would indicate that the network of the primary variable is neither sufficient nor necessary and would provide a model system for the study of the aging determinants other than those associated with the controlled variable. Significant attributes of complex systems can rarely be captured by single variables but require composite multivariate measures. For example, Galsworthy and colleagues (2002) described a first principal component of variables from six cognitive tasks to characterize general learning ability in HS mice. Other multivariate indices with attractive prospects include canonical correlations that provide the composite of measures with the highest stability across longitudinal measurements, discriminant functions that maximally distinguish between specific genotypes, and cluster analyses assigning individuals to groupings based on several phenotypic measures. Genotypic Selective Breeding Genotypic selective breeding involves the mating of animals with specific alleles of loci known or hypothesized to affect the phenotype to produce research models according to genotypic specification. In contrast to models generated by phenotypic selection, which involve the entire complex network of mechanisms in contrasting configurations, genotypic selection generates groups that are distinguished by only those aspects ILAR Journal of the mechanistic network that are downstream from the specific selected locus or loci, thus enabling focused evaluation of a subsection of that network. Genotypic selection has long been used in the development of congenic strains, through the placement of a particular gene or chromosomal region (or combinations of such genes) by successive backcrossing on differing inbred strain backgrounds. Advances in molecular genetics now provide an unprecedented roster of candidates for this procedure and QTL methods enhance this capability by encompassing the loci of a known chromosomal location but unknown molecular identity. This method was promoted by Harrison and colleagues (1995) under the concept of “allele capture by selection for flanking markers,” and Bennett and Johnson (1998) used the application to assess QTLs for hypnotic sensitivity to ethanol to generate “speed congenics.” Genotypic selection is also effective for placing specific genetic material on a heterogeneous background. Examples of this usage include the confirmation or rejection of QTLs nominated for influence on alcohol-related processes in mice (Geraghty 1998; Tarantino et al. 1998). Genotypic selection also offers a powerful capability to assemble panels of genes and/or QTLs in any specifiable combination to study the interactions of the constituent elements of a complex determinative system. Advantages of genotypic selection over phenotypic selection in this context are the availability of results (for one or a few genes or QTLs) in a single generation and the ability to reconstitute the lines at any time. The lines require maintenance by consecutive-generation breeding only as long as necessary for a particular research purpose. If the need reappears after termination of the lines, or in another laboratory, the lines can be assembled in their specifiable configuration within a few generations. In contrast, with conventional selection a terminated line is gone forever. Models generated by genotypic selection will not replace conventional phenotypically selected lines but will be a powerful complement to them. It is therefore easy to predict that genotypic selection will become a primary method for the tailoring of animal models for gerontological research. characters at an early age and relating them to the late-life measures of interest. For example, Luckinbill and Clare (1985) and Rose and Charlesworth (1981) generated extensively studied Drosophila lines by selecting for late fecundity. A key to success with this strategy is accurate identification of biomarker variables of youth that have predictive value for late-life phenomena. Numerous measures have been thus identified, but developmental considerations suggest that such long-term effects cannot be assumed. An approach with multivariate indices might provide maximum predictive value from early-life measurements. One generation of animals could provide from a panel of early life measurements the multivariate index that maximally predicts lifespan, which could then be the selected phenotype. Selective Breeding and Lifespan Variables References The research strategy in generating tools to specification by phenotypic selective breeding is to breed for line differences in the target phenotype, and compare the resulting lines for “correlated characters” hypothesized as mediating, intervening, and mechanistic variables in the genome, proteome, or metabolome. But there is a challenge to the use of this approach in gerontological and healthspan research: many of the most relevant phenotypes are measurable only in late life or at the end of life, past the possibility of reproduction of the selected animals. Harrison and Roderick (1997) suggested one solution: a form of family selection, mating individuals on the basis of the lifespans of their grandparents. Temporal constraints suggest the utility of selecting directly for one or more correlated (and presumptive causal) Austad SN. 1993. The comparative perspective and choice of animal models in aging research. Aging Clin Exp Res 5:259-267. Bennett B, Johnson TE. 1998. Development of congenics for hypnotic sensitivity to ethanol by QTL-marker-assisted counter selection. Mamm Genome 9:969-974. Blizard DA, Lionikas A, Vandenbergh DJ, Vasilopoulos T, Gerhard GS, Klein LC, Stout JT, Mack HA, Lakoski JM, Larsson L, Spicer JM, Vogler GP, McClearn GE. 2009. Blood pressure and heart rate QTL in mice of the B6/D2 lineage: Sex differences and environmental influences. Physiol Genomics 36:158-166. Burke MK, Rose MR. 2009. Experimental evolution with Drosophila. Am J Physiol Regul Integr Comp Physiol 296:R1847-R1854. Chesler EJ, Miller RA, Branstetter LR, Galloway LD, Jackson BL, Philip VM, Voy BH, Culiat CT, Threadgill DW, Williams RW, Churchill GA, Johnson DK, Manly KF. 2008. The collaborative cross at Oak Ridge National Laboratory: Developing a powerful resource for systems genetics. Mamm Genome 19:382-389. Volume 52, Number 1 2011 Summary It is clear that there is no one “best” model (Sprott 1997) and that the optimal model or combination of models depends on the circumstances. The diversity and efficacy of the research tools provided by the genetic sciences offer underexploited and powerful opportunities. Heterogeneous stocks and selective breeding permit the assembly of model systems that: (1) scramble alleles (i.e., those that are polymorphic in the founding parent strains) in order to provide normative groups, vehicles for assessment of complex correlational relationships, and foundation stocks for selective breeding; (2) concentrate alleles that are relevant to a particular phenotype into contrasting groups of animals—one with all of the increasing alleles and one with all of the decreasing alleles—in order to provide material for the investigation of the total mechanistic network through which the phenotypic differential is accomplished; and (3) concentrate configurations of alleles of specific loci (or QTLs) in order to permit examination of the dynamics of a subset of the mechanism that is causally downstream from those loci. 21 Chia R, Achilli F, Festing MFW, Fisher EMC. 2005. The origins and uses of mouse outbred stocks. Nat Genet 37:1181-1186. Chrisp CE, Turke P, Luciano S, Swalwell S, Peterson J, Miller RA. 1996. Lifespan and lesions in genetically heterogeneous (four-way cross) mice: A new model for aging research. Vet Pathol 33:735-743. Churchill GA, Airey DC, Allayee H, Angel JM, Attie AD, Beatty J, Beavis WD, Belknap JK, Bennett B, Berrettini W, Bleich A, Bogue M, Broman KW, Buck KJ, Buckler E, Burmeister M, Chesler EJ, Cheverud JM, Clapcote S, Cook MN, Cox RD, Crabbe JC, Crusio WE, Darvasi A, Deschepper CF, Doerge RW, Farber CR, Forejt J, Gaile D, Garlow SJ, Geiger H, Gershenfeld H, Gordon T, Gu J, Gu W, de Haan G, Hayes NL, Heller C, Himmelbauer H, Hitzemann R, Hunter K, Hsu HC, Iraqi FA, Ivandic B, Jacob HJ, Jansen RC, Jepsen KJ, Johnson DK, Johnson TE, Kempermann G, Kendziorski C, Kotb M, Kooy RF, Llamas B, Lammert F, Lasalle JM, Lowenstein PR, Lu L, Lusis A, Manly KF, Macucio R, Matthews D, Medrano JF, Miller DR, Mittleman G, Mock BA, Mogil JS, Montagutelli X, Morahan G, Mott R, Nadeau JH, Nagase H, Nowakowski RS, O’Hara BF, Osadchuk AV, Page GP, Paigen K, Palmer AA, Pan HJ, Petonen-Palotie L, Peirce J, Pomp D, Pravenec M, Prows DR, Qi Z, Reeves RH, Roder J, Rosen GD, Schadt EE, Schalkwyk LC, Seltzer Z, Shimomura K, Shou S, Sillanpaa MJ, Siracusa LD, Snoeck HW, Spearow JL, Svenson K, Tarantino LM, Threadgill DW, Toth LA, Valdar WS, deVillena FP, Warden C, Whatley S, Williams RW, Wiltshire T, Yi N, Zhang D, Zhang M, Zou F. 2004. The collaborative cross, a community resource for the genetic analysis of complex traits. Nat Genet 36:1133-1137. Falconer DS. 1992. Early selection experiments. Annu Rev Genetics 26:1-14. Falconer DS, Mackay TF. 1996. Introduction to Quantitative Genetics, 4th ed. Essex: Longman Group. Fernandez-Teruel A, Escorihuela RM, Gray JA, Aguilar R, Gimenez-Llort L, Tobena A, Bhomra A, Nicod A, Mott R. 2002. A quantitative trait locus influencing anxiety in the laboratory rat. Genome Res 12:618-626. Festing MFW. 1999. Warning: The use of heterogeneous mice may seriously damage your research. Neurobiol Aging 20:237-244. Fosmire GJ, Focht SJ, McClearn GE. 1993. Genetic influences on tissue deposition of aluminum in mice. Biolog Trace Element Res 37:115-121. Galsworthy MJ, Paya-Cano JL, Monleon S, Plomin R. 2002. Evidence for general cognitive ability (g) in heterogeneous stock mice and an analysis of potential confounds. Genes Brain Behav 1:88-95. Galsworthy MJ, Paya-Cano JL, Liu L, Monleon S, Gregoryan G, Fernandes C, Schalkwyk LC, Plomin R. 2005. Assessing reliability, heritability and general cognitive ability in a battery of cognitive tasks for laboratory mice. Behav Genet 35:675-692. Gelman R, Watson A, Bronson R, Yunis E. 1988. Murine chromosomal regions correlated with longevity. Genetics 118:693-704. Geraghty JT. 1998. A review of the genetics of alcoholism and a confirmatory study of an acute alcohol withdrawal quantitative trait locus in mice. In: Intercollege Graduate Program in Genetics: Master of Science. University Park: Pennsylvania State University. Harrison DE, Roderick TH. 1997. Selection for maximum longevity in mice. Exp Gerontol 32:65-78. Harrison DE, Roderick TH, Paigen K. 1995. Allele capture by selection for flanking markers: A new method for analyzing multigenic traits. Growth Dev Aging 59:73-76. Harrison DE, Strong R, Sharp ZD, Nelson JF, Astle CM, Flurkey K, Nadon NL, Wilkinson E, Frenkel K, Carter CS, Pahor M, Javors MA, Fernandez E, Miller RA. 2009. Rapamycin fed late in life extends lifespan in genetically heterogeneous mice. Nature 460:392-395. Heller DA, Ahern FM, Stout JT, McClearn GE. 1998. Mortality and biomarkers of aging in heterogeneous stock (HS) mice. J Gerontol 53A:B217B230. Huang G-J, Shifman S, Valdar WS, Johannesson M, Yalcin B, Taylor MS, Taylor JM, Mott R, Flint J. 2009. High resolution mapping of expression QTLs in heterogeneous stock mice in multiple tissues. Genome Res 19:1133-1140. Kaneko K. 2009. Relationship among phenotypic plasticity, phenotypic fluctuations, robustness, and evolvability: Waddington’s legacy revisited under the spirit of Einstein. J Biosci 34:529-542. 22 Lang DH, Sharkey NA, Mack HA, Vogler GP, Vandenbergh DJ, Blizard DA, Stout JT, McClearn GE. 2005. Quantitative trait loci analysis of structural and material skeletal phenotypes in C57BL/6J and DBA/2 secondgeneration and recombinant inbred mice. J Bone Miner Res 20:88-99. Lang DH, Gerhard GS, Griffith JW, Vogler GP, Vandenbergh DJ, Blizard DA, Stout JT, Lakoski JM, McClearn GE. 2010. Quantitative trait loci (QTL) analysis of longevity in C57BL/6J by DBA/2J (BXD) recombinant inbred mice. Aging Clin Exp Res 22:8-19. Leiter EH, Coleman DL, Hummel KP. 1981. The influence of genetic background on the expression of mutations at the diabetes locus in the mouse. III. Effect of H-2 haplotype and sex. Diabetes 30:1029-1034. Liao CY, Rikke BA, Johnson TE, Diaz V, Nelson JF. 2010. Genetic variation in the murine lifespan response to dietary restriction: From life extension to life shortening. Aging Cell 9:92-95. Lionikas A, Blizard DA, Vandenbergh DJ, Stout JT, Vogler GP, McClearn GE, Larsson L. 2006. Genetic determinants of weight of fast- and slowtwitch skeletal muscles in old mice. Mamm Genome 17:689-699. Luckinbill LS, Clare MJ. 1985. Selection for life span in Drosophila melanogaster. Heredity 55:9-18. McClearn GE, Hofer SM. 1999. Genes as gerontological variables: Genetically heterogeneous stocks and complex systems. Neurobiol Aging 20:147-156. McClearn GE, Kakihana R. 1981. Selective breeding for ethanol sensitivity: SS and LS mice. In: McClearn GE, Deitrich RA, Erwin VG, eds. Development of Animal Models as Pharmacogenetic Tools. DHEW Publication No. [ADM] 81-1133. Washington: US Government Printing Office. p 147-159. McClearn GE, Meredith W. 1964. Dimensional analysis of activity and elimination in a genetically heterogeneous group of mice (Mus musculus). Anim Behav 12:1-10. McClearn GE, Wilson JR, Meredith W. 1970. The use of isogenic and heterogenic mouse stocks in behavioral research. In: Lindzey G, Thiessen DD, eds. Contributions to Behavior-Genetic Analysis: The Mouse as Prototype. New York: Appleton-Century-Crofts. p 3-22. McClearn GE, Deitrich RA, Erwin VG. 1981. Development of Animal Models as Pharmacogenetic Tools. Research Monograph No. 6, DHEW Publication No. [ADM] 81-1133. Washington: US Government Printing Office. Miller RA, Chrisp C, Jackson AU, Burke D. 1998. Marker loci associated with life span in genetically heterogeneous mice. J Gerontol 53A:M257M263. Miller RA, Austad SN, Burke D, Chrisp C, Dysko R, Galecki A, Jackson AU, Monnier V. 1999a. Exotic mice as models for aging research: Polemic and prospectus. Neurobiol Aging 20:217-231. Miller RA, Burke D, Nadon NL. 1999b. Four-way cross mouse stocks: A new, genetically heterogeneous resource for aging research. J Gerontol 54A:B358-B360. Miller RA, Harper JM, Galecki A, Burke D. 2002. Big mice die young: Early life body weight predicts longevity in genetically heterogeneous mice. Aging Cell 1:22-29. Minelli A, Fusco G. 2010. Developmental plasticity and the evolution of animal complex life. Philos Trans R Soc Lond B Biol Sci 365:631-640. Petkov PM, Graber JH, Churchill GA, DiPetrillo K, King BL, Paigen K. 2005. Evidence of a large-scale functional organization of mammalian chromosomes. PLoS Genetics 1:e33. Phelan JP, Austad SN. 1994. Selecting animal models of human aging: Inbred strains often exhibit less biological uniformity than F1 hybrids. J Gerontol 49:B1-B11. Roberts RC. 1967. Some concepts and methods in quantitative genetics. In: Hirsch J, ed. Behavior-Genetic Analysis. New York: McGraw-Hill. p 214-257. Rose MR, Charlesworth B. 1981. Genetics of life history in Drosophila melanogaster. II. Exploratory selection experiments. Genetics 97:187196. Rose MR, Rauser CL, Mueller LD, Benford G. 2006. A revolution for aging research. Biogerontology 7:269-277. Silver LM. 1995. Mouse Genetics: Concepts and Applications. New York: Oxford University Press. ILAR Journal Sprott RL. 1997. Mouse and rat genotype choices. Exp Gerontol 32:79-86. Strauss JA, Ballough GP, Anthony A, Mitchell RB. 1992. Changes in HS mouse erythrocyte mass and area with respect to donor age. J Gerontol 47:B13-B17. Swindell WR, Harper JM, Miller RA. 2008. How long will my mouse live? Machine learning approaches for prediction of mouse life span. J Gerontol 63A:895-906. Takeda T. 2009. Senescence-accelerated mouse (SAM) with special references to neurodegeneration models, SAMP8 and SAMP10 mice. Neurochem Res 34: 639-659. Talbot CJ, Radcliffe RA, Fullerton J, Hitzemann R, Wehner JM, Flint J. 2003. Fine scale mapping of a genetic locus for conditioned fear. Mamm Genome 14:223-230. Tarantino LM, McClearn GE, Rodriguez LA, Plomin R. 1998. Confirmation of quantitative trait loci for alcohol preference in mice. Alcohol Clin Exp Res 22:1099-1105. Threadgill D, Miller D, Churchill GA, Pardo-Manuel de Villena F. 2011. The Collaborative Cross: Recombinant inbred lines in the systems genetic era. ILAR J 52:24-31. Turri MG, Henderson ND, DeFries JC, Flint J. 2001. Quantitative trait locus mapping in laboratory mice derived from a replicated selection experiment for open-field activity. Genetics 158:1217-1226. Volume 52, Number 1 2011 Valdar WS, Flint J, Mott R. 2003. QTL fine-mapping with recombinantinbred heterogeneous stocks and in vitro heterogeneous stocks. Mamm Genome 14:830-838. Valdar WS, Solberg LC, Gauguier D, Burnett S, Klenerman P, Cookson WO, Taylor MS, Rawlins JN, Mott R, Flint J. 2006a. Genome-wide genetic association of complex traits in heterogeneous stock mice. Nat Genet 38:879-887. Valdar WS, Solberg LC, Gauguier D, Cookson WO, Rawlins JN, Mott R, Flint J. 2006b. Genetic and environmental effects on complex traits in mice. Genetics 174:959-984. Vieira C, Pasyukova EG, Zeng ZB, Hackett JB, Lyman RF, Mackay TF. 2000. Genotype-environment interaction for quantitative trait loci affecting life span in Drosophila melanogaster. Genetics 154:213227. Waddington CH. 1956. Genetic assimilation of the bithorax phenotype. Evolution 10:1-13. Williams RW, Bennett B, Lu L, Gu J, DeFries JC, Carosone-Link PJ, Rikke BA, Belknap JK, Johnson TE. 2004. Genetic structure of the LXS panel of recombinant inbred mouse strains: A powerful resource for complex trait analysis. Mamm Genome 15:637-647. Yuan R, Peters LL, Paigen B. 2011. Mice as a mammalian model for research on the genetics of aging. ILAR J 52:4-15. 23