Download Heterogeneous Reference Populations in Animal

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Epistasis wikipedia , lookup

Site-specific recombinase technology wikipedia , lookup

Human–animal hybrid wikipedia , lookup

Inbreeding avoidance wikipedia , lookup

Twin study wikipedia , lookup

Designer baby wikipedia , lookup

Genetically modified food wikipedia , lookup

Pharmacogenomics wikipedia , lookup

Genetic testing wikipedia , lookup

Hardy–Weinberg principle wikipedia , lookup

Pathogenomics wikipedia , lookup

Polymorphism (biology) wikipedia , lookup

Dominance (genetics) wikipedia , lookup

Human genetic variation wikipedia , lookup

Public health genomics wikipedia , lookup

Genome (book) wikipedia , lookup

Genetic drift wikipedia , lookup

Medical genetics wikipedia , lookup

Behavioural genetics wikipedia , lookup

Quantitative trait locus wikipedia , lookup

Heritability of IQ wikipedia , lookup

Koinophilia wikipedia , lookup

Inbreeding wikipedia , lookup

Genetic engineering wikipedia , lookup

History of genetic engineering wikipedia , lookup

Genetic engineering in science fiction wikipedia , lookup

Population genetics wikipedia , lookup

Microevolution wikipedia , lookup

Transcript
Heterogeneous Reference Populations in Animal Model
Research in Aging
Gerald E. McClearn
INTRODUCTION
For many biological disciplines, such as ecology, wild populations are the primary object of study. In other research endeavors, comparative studies examine animals of different species, either wild-trapped or laboratory-reared, and either in
their natural habitat or in laboratory settings. Such comparative observations have played a very substantial role in gerontological science. The monumental study of Finch (1990)
has provided a summary of this work. Advantages of the
comparative approach have been succinctly argued by Austad
(1993), who observes, with regret, that contemporary gerontology has "lost its comparative focus." Central among the
themes he pursues is the limited perspective provided by rodent models that have provided so much of laboratory-derived
data in gerontology. An underlying concern is whether the
model provided by any given species will provide results that
will generalize to humankind. This paper is concerned not
with these broad and major issues, but with the more narrow
one of generality to the species as a whole of conclusions
obtained from typical laboratory observations on genotypically restricted members of that species.
GENES AS VARIABLES
In general, in animal model research, the basic motivation
for comparing observations from laboratories to observations on "wild" populations is to determine whether controlled variables of the laboratory setting have produced an
outcome that is generalizable to the real world. It is an
interesting philosophical problem. The laboratory permits
control that simplifies complex systems so that relationships
between and among variables may be perceived. In effect,
the laboratory study takes a small piece of the total causal
field, isolates it as far as possible from the rest of that field,
and evaluates the extent to which a change in variable X is
associated with a change in variable Y. The highly controlled conditions may, however, constrict the range of circumstances under which the observed relationship between
X and Y will obtain, and the issue is whether that relation-
ship can be shown when controlled variables are unfettered.
Such unfettering can be accomplished in a systematic way in
a laboratory setting, of course, but the total immersion tactic
of testing in a real world setting is sometimes championed.
The process described above applies to both environmental and genetic variables. Dealing with environmental
variables is at the heart of all biological disciplinary specialties, and acquisition of knowledge about identification, manipulation, control, and evaluation of these variables constitutes much of the education of the biologist. Dealing
appropriately with nutritional status of the diet, caging density, light cycles, time of day of testing, dose administration,
procedures for handling, frequency of cage changing, control of high frequency noise—all of these and countless more
similar variables—is a mark of good practice in research in a
disciplinary area.
It is interesting to contemplate that the more variables
that must be controlled for a relationship of X to Y to be
shown reliably, the less proportional impact X must have in
the more complex, uncontrolled, real world situation. Observations in a reference situation with relaxed environmental controls can determine the generality-restricting consequences exerted by controls imposed during the conduct of
research.
There are also genetic constraints in laboratory research.
For some biological disciplines, these are the core subject
matter. In others, however, genetic constraints are less well
understood and appreciated than are the environmental ones.
The present discussion will focus, therefore, on the logic of
genetic reference groups, which relax some of the genetic
controls typical of laboratory research. The notion of genetic
context is of central importance. The basic point is that the
influence of allelic differences at a particular locus may depend on the alleles present at other loci of the organism. A
particularly apposite example of this type of interaction is
that of Coleman and Hummel (1975), who showed that animals homozygous for an allele "causing" diabetes could display dramatically different symptom patterns depending on
background genotype.
MANIPULATION AND CONTROL OF
GENES
Gerald McClearn, Ph.D., is Director of the Center for Developmental and
Health Genetics and Evan Pugh Professor of Health and Human Development at the Pennsylvania State University, University Park, Pennsylvania.
Volume 38, Number 3
1997
It is useful to consider genes as controlled or as independent
variables. Until very recently, manipulation of genotype
119
could be accomplished only by indirect means, through assignment of mates. Strong theories of Mendelian genetics,
cytogenetics, population genetics, and quantitative genetics
were relied on to deduce the genotypic consequences of a
particular mating program. The most utilized of these mating
programs for the generation of animal models have been
inbreeding and phenotypic selective breeding, in which the
genotypes are indirectly rearranged. Direct manipulations of
genetic engineering, which can be expected to become increasingly important in gerontological research as elsewhere,
will become parallel methods that will build on animal groups
generated by these indirect methods but will not replace
them.
Inbred Strains and Derived Generations
Inbreeding is the great genetic homogenizing technique. The
offspring of matings of relatives are more likely to be homozygous at any particular genetic locus than offspring of
unrelated parents. If inbreeding is continued in successive
generations, the homozygosity increases, and in any line of
descent so maintained, homozygosity becomes nearly total.
As a close approximation, it can be assumed that all animals
within an inbred strain are homozygous in like state at all
loci. Homozygosity is closely approached after 20 consecutive generations in which each generation is derived from a
single sibling pair. If the mating scheme is continued, this
condition of genetic uniformity will be maintained, aside
from the possibility of mutations. Inbred strains are thus quite
stable and constitute the principal means of genetic control
in current use. In many cases, this control is exerted simply
by employing a single strain. Within a given laboratory over
time, and in different laboratories, investigators using that
particular strain are confident that they are dealing with the
same basic biological system. Borrowing terms from the
language of environmental control, the genotype of an inbred strain might be said to be fixed, or "controlled," at a
single "level." The level might be better termed a configuration, because in most cases, there is no obvious metric that
will permit description of the system of anonymous polygenes in terms of unidimensional levels. It is also important
to note that the genotype of a particular inbred strain is
achieved not by a directional process, but by a stochastic
one. Thus, an inbred strain is not bred "for" some phenotype; it is simply inbred, and nothing can be inferred about
the genotype except that it has approached the condition of
being homozygous at all loci.
Comparisons among several inbred strains will often
identify one displaying a particularly desirable level of some
phenotype of interest. That strain can then constitute a model
system for the study of the phenotype in detail. For example,
C57BL/6 mice are outstanding among the more commonly
used mouse strains for their voluntary alcohol consumption
when offered a choice between alcohol solution and water
(McClearn and Rodgers 1959); they have become, there-
120
fore, the strain of choice for any research requiring animals
to consume some alcohol. The gradually accumulating
knowledge about individual strains makes them more and
more desirable as new research results can be referred to an
ever-expanding data base.
A frequently employed research tactic is to employ several strains with different levels of the phenotype. From one
perspective, use of multiple strains is an approximation to
control by fixation at different levels. From another perspective, multiple strain comparisons can be thought of as manipulation of genotype as an independent variable, because
choice of strain is at the discretion of the investigator. This
is a quite limited manipulation, however. Strictly speaking,
for genotype to be an independent variable in the classical
sense, it should be possible for the investigator to assign a
gene or a set of specified genes to animals at random, in the
same way that drug doses or degrees of stress or caloric
restriction can be assigned. This is obviously impossible,
and the investigator must settle for a choice of whole, anonymous, and mostly unknown genotypes.
This use of inbred strains of mice and of rats has long
been a popular approach to the study of aging. Logistical
difficulties encountered by individual investigators in maintaining colonies of aged animals of different strains under
appropriate environmental conditions eventuated in a formal
program of the National Institute on Aging to provide animals to meet the need of investigators in the field. In turn,
availability of these animals has accentuated demand, so that
the fixed-genotype logic has become a major component of
animal model gerontology.
The major strength of the inbred strain approach, as
already noted, derives from their uniformity and relative
stability over time. This uniformity is also the principal
weakness. For example, the only source for variability
within an inbred strain is the environment, broadly defined.
Any attempt to describe the relationship between 2 phenotypes within a strain will be constrained by the fact that
variances, and therefore covariances of measured phenotypes, are exclusively of environmental origin; genetic influences cannot be detected because there are none present.
Of more fundamental import, inbred strains are very limited as representatives of their species. Each strain represents only 1 of an indefinitely large number of genotypes
that can be generated from the parent population. Furthermore, it is not as though a single normal genotype was
sampled randomly and then indefinitely replicated in an
inbred strain. Members of a strain are homozygous at all
loci and can best be considered as a single gamete that is
duplicated and replicated. Such a genetic configuration
could never occur in a bisexually reproducing natural population. Even worse, the gamete thus duplicated cannot be
considered a representative sample from all possible gametes from the population. In the population at large, gametes that have an allele that is lethal when homozygous
may persist in heterozygotes. Because of the homozygosity achieved by inbreeding, however, these alleles will be
ILAR Journal
eliminated. Any single inbred strain thus provides only a
very limited sampling of the gene pool of its species.
Along with the inbred strains, the hybrid Fl animals obtained by crossing different strains have found frequent employment in research on aging. A degree of manipulation of
genotype in the classical sense is accomplished by crossing
inbred strains, because the resulting Fl hybrid has known, or
very strongly inferred, genotype characteristics relative to
the 2 parent strains: For each locus for which the 2 strains
have different alleles, the Fl animals will be heterozygotes.
Again, random assignment is not possible; the investigator
cannot pick an animal and assign it to be a homozygote or
heterozygote, but it can be said that all Fl animals will have
an allelic dose halfway between those of the parents. It is
worth noting that this firm positioning of the Fl is relative to
2 mostly or completely unknown anchor points, because the
nature and number of such alleles is unknown. A geographic
analogy would be knowing that some place is exactly halfway between 2 other landmarks but not knowing the location
of either of them. In spite of this uncertainty, the relationship
of Fl to parent strains is quite useful in that it permits the
most elementary of the biometric operations of quantitative
genetics. In short, it becomes possible to relate the allelic
dose to the phenotype.
Fl animals, unlike their inbred parents, will exhibit much
heterozygosity. Indeed, they will be heterozygous at all loci
for which the parents had different alleles. However, all Fl
animals are genetically alike, and none will be homozygous
at any of these segregating loci. In this respect, the Fls are
no better representatives of the species than inbred strains.
Subsequent mating operations can further this type of
analysis substantially. An F2 obtained by mating Fl animals, unlike their Fl parents or their inbred strain grandparents, comprises individuals that differ genetically from each
other. As a consequence of the reshuffling of the parent
strain alleles, each genotype of each locus occurs in a variable genetic context of the genotypes at all other loci. Insofar as the effect of genotypic differences at any one locus is
dependent to some extent on the genotypes at other loci, on
the heterogeneous background, this effect appears as a sort
of average, more representative of the operation of the gene
locus than would be evidenced on any single inbred background.
In the case of F2 or other intercross groups such as
backcrosses, the manipulation of genotype by assignment of
mates of particular types has predictable consequences for
the group of offspring but is not informative about particular
individuals in the group. Analytic attention is thus focused
on the variability within the group, and the objective becomes that of ascribing the measured phenotypic variance to
genetic and environmental influences, to their covariances,
and to their interactions. The F2 and backcrosses provide a
reference context within which the inbred and Fl results can
be interpreted. An elementary example can be derived from
observations on 2 inbred strains and their Fl and F2 generations. The variance of the phenotype of the inbred parental
Volume 38, Number 3
1997
strains and of the Fl is exclusively environmental in origin;
that of the F2 has both environmental and genetic components. Subtracting the former from the latter provides an
estimate of genetic variance, and these can be formed into a
ratio of genetic to total phenotypic variance to generate a
simple index of "heritability." The field of quantitative genetics is a highly developed conceptual and analytical model
for dealing with this type of situation. Falconer and MacKay
(1996) provide one of several excellent introductions to the
area, and Phelan (1992) has related these issues to aging.
Phenotypic Selective Breeding
Selective breeding is directional. Beginning with a genetically heterogeneous foundation stock (see below), the investigator mates animals of like extreme phenotypic value. If
heritability is nonzero, there will be a gradual accumulation
of alleles resulting in high phenotypic scores in one line and
low scores in the other line. Phenotypic selection is a comprehensive procedure, in that an ideally executed study will
locate and parcel out all the structural and regulatory genes
segregating in the gene pool of the foundation stock that
have any influence on the phenotype in question. It remains
the case, however, that genes sorted out in this way are
anonymous.
Although there are limited circumstances under which
heterozygosity might be maintained in the selected loci, the
general expectation is that the "high" line will become homozygous for the "increasing" alleles and the "low" line will
become homozygous for the "decreasing" alleles. This homozygosity presents a substantially different picture from
that of inbred strains, however. First, as noted, selection
provides for a relative uniformity of direction of effect of the
alleles within a line. All, or most, of the relevant loci in a
selected line will be homozygous for the increasing (or decreasing) alleles, whereas in inbred strains, the configuration
is fortuitous, with both increasing and decreasing alleles in
differing proportions in the different strains. Furthermore,
unlike the case in inbred strains, the nonrelevant genotype is
heterogeneous. This feature offers clear opportunities for
assessing hypotheses about correlated characteristics. For
example, C57BL/6 inbred mice are widely reported to have
longer lives than DBA/2 mice. It is tempting to test theoretical expectations (about oxidative stress, for illustration) by
comparing appropriate measures (such as particular enzyme
activity levels) of the 2 strains. Results from such a comparison are very weak, however. The stochastic process of inbreeding will have fixed all loci, including those pertinent to
the enzymes in question. Thus, even if no relationship exists
between a given enzyme activity and longevity, inbreeding
will provide 4 possible outcomes with respect to the enzyme
genotype (assume single locus inheritance for the enzyme
activity): both strains with high enzyme activity, both strains
with low enzyme activity, C57BL/6s with high and DBA/2s
with low activity, and C57BL/6s with low and DBA/2s with
121
high activity. One quarter of the outcomes would falsely
appear to confirm the hypothesis. The situation can be improved by employing multiple strains instead of only 2; however, the power of the analysis is limited by the number of
strain means, not by the number of individuals, and the approach may not be very cost-effective compared with the
alternative tactics offered by selected lines or by heterogeneous stocks.
Selectively bred lines, by virtue of a more-or-less randomly segregating "residual" genotype (that genotype not
relevant to the selected attribute), offer a particularly powerful means of evaluating hypotheses about correlated characteristics. Within the usual limits of sampling considerations,
any trait showing a significant difference between the 2 selected lines can be regarded as being functionally involved in
the causal nexus of the phenotype for which selection was
performed. Because in the real world of finite breeding population size there will always be the possibility of sampling
bias, replicate selected lines are frequently employed. A
consistent outcome in the replicates provides compelling evidence for the relationship.
One limitation of selection, however, is that the genetic
array relevant to the selected phenotype, and partitioned in
the upward and downward selected lines, will be limited to
the allelic differences that were available in the foundation
population. Another selection program for the same phenotype, but beginning with a different gene pool, might result
in the same phenotypic difference, but for somewhat, substantially, or entirely different genetic reasons. Nevertheless, selective breeding procedures have proved valuable in
aging research on Drosophila (see Arking [1991] and Rose
[1991] for review of these studies) and are in active progress
in application to mammals (Harrison and others 1995).
Heterogeneous Stocks
If the value of heterogeneous stocks is to be found in the
extent to which they represent their entire species, it would
seem that the optimal strategy would be to sample from the
"real" world of mice, rats, or whatever species is involved.
Certainly, the value of a genetic context will be a positive
function of how varied it is. In other words, it will be advantageous to have as much allelic variation as possible. For
example, it would be ideal to have in one's reference group
every existing allele in all strains of mice; however, this ideal
is not attainable for several reasons. In the first instance, no
practical trapping program could sample all the world's mice.
Any local trapping will be sampling a unique gene pool and
will be no more representative of every strain of mice than
sampling a village will provide all the genetic variation of
humankind. Furthermore, trapping biases are well-known;
trap-shy mice will be omitted, along with any of their genes
that result in trap-shyness. Upon importation to the laboratory, these wild-trapped animals will suddenly encounter a
stupendously different environment from their accustomed
one, and natural selection will begin forthwith to weed out
122
those alleles from the wild that lead to reduced survival and
fertility in the laboratory.
In general, then, establishing heterogeneous stocks by
wild-trapping would have the very great advantage of introducing new alleles that have been absent from the existing
stocks with their idiosyncratic origins. However, these wildderived stocks would become less and less representative of
the wild population as they became laboratory-adapted. This
author argues that the issue of such adaptation is not really
important. Indeed, for today's laboratory animals, after generations of such selection, the laboratory is the natural habitat, and the heterogeneous reference populations should be
adapted to laboratories, not to fields and forests.
Another route to establishing heterogeneous reference
stocks does not provide for the infusion of new genetic material to the laboratory stocks but does shuffle what is already
present. This route involves the generation of genetically
segregating populations by intercrossing of inbred strains.
Such an approach has already been hinted, with the observation that an F2 consists of genetically unique animals. It is
pertinent to ask just how genetically variable an F2 really is.
The answer must be stated in reference to the parent strains
from which the Fl generation was derived, because genetic
variability in the F2 will depend on the number of loci from
which those parents differed in allelic state and for which
specific alleles they differed. Thus, some F2s will be more
genetically variable than others. It should be noted, however, that no F2 can offer more than 2 alleles per locus,
because there are only 2 progenitor strains and each of them
is homozygous for a single allele at each locus. Thus, although it provides scope for expression of genes against a
somewhat heterogeneous genetic milieu, this milieu will be
idiosyncratic to the particular progenitor base.
If an objective is to sample more completely from the
totality of genetic information available within a species,
then an obvious move is to expand this progenitor base. For
example, a 4-way cross can be generated by mating an Fl
between 2 progenitors with another Fl from different progenitors. There is now scope for more segregating loci, and
the sampling limit is now 4 possible alleles. Such expansion
can proceed even further with 8-way or more comprehensive
intercrosses, with advantages accruing from each increment
in number of alleles per locus. The logic of using heterogeneous stocks to provide foundation stocks has been discussed
in some detail elsewhere (McClearn 1979; McClearn and
others 1970). Briefly, such stocks are particularly valuable
when used (1) as foundation stocks for selective breeding,
(2) to provide reference norms for the evaluation of means
and variances of inbred strains, (3) to examine correlations
between variables, or (4) to evaluate the generality of a genetic effect identified in groups with constrained genetic variability.
Current use of standardized heterogeneous stocks in
gerontological research is infrequent, but awareness of their
advantages is growing. McClearn (1992a,b), for example,
has described the use of genetically heterogeneous (HS)
ILAR Journal
mice, derived from 8 inbred strains, both in evaluating the
heterogeneity and reliability of a panel of behavioral
biomarkers of aging and in exploring the structural stability
and change reflected in the patterns of relationships among
the variables over the lifespan. More recently, a 4-way cross
population of mice has been used to test hypotheses about
interrelationships of various immune system parameters and
their relationships to lifespan and cancer incidence (Miller
and others 1994), to assess variation in lifespan and in lesions (Chrisp and others 1996), and to validate candidate
biomarkers of aging reflecting immune system status and
muscle function (Miller and others 1997).
These few examples illustrate the many desirable attributes of genetically heterogeneous animal models in
gerontological research. As these attributes are increasingly
appreciated, it is likely that standardized heterogeneous
stocks will eventually be recognized as essential complements to the inbred strains and derived generations of current
popularity.
REFERENCES
Arking R. 1991. Biology of Aging: Observations and Principles. Englewood
Cliffs NJ: Prentice-Hall.
Austad SN. 1993. The comparative perspective and choice of animal
models in aging research. Aging Clin Exp Res 5:259-267.
Chrisp CE, Turke P, Luciano A, Swalwell S, Petersen 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.
Coleman DL, Hummel KP. 1975. Influence of genetic background on the
expression of mutations at the diabetes locus in the mouse. II. Studies
on background modifiers. Israeli J Med Sci 11:708-713.
Falconer DS, Mackay TRC. 1996. Introduction to Quantitative Genetics.
4th ed. Essex, England: Longman.
Volume 38, Number 3
1997
Finch CE. 1990. Longevity, Senescence, and the Genome. Chicago: The
University of Chicago Press.
Harrison DE, Roderick TH, Paigen K. 1995. Allele capture by selection for
flanking markers: A new method for analyzing multigenic traits. Growth
Devel Aging 59:73-76.
McClearn GE. 1979. Influence of genetic variables on means, variances,
and covariances in behavioral responses to toxicological and pharmacological substances. J Toxicol Environ HI 5:145-156.
McClearn GE. 1992a. The reliability and stability of biomarkers of aging.
Ann NY Acad Sci 673:1-8.
McClearn GE. 1992b. Heterogeneity of biomarkers of aging. In: Licastro
F, Caldarera CM, editors. Biomarkers of Aging: Expression and Regulation. Bologna: Editirice Bologna, p 27-36.
McClearn GE, Rodgers DA. 1959. Differences in alcohol preference among
inbred strains of mice. Q J Stud Alcohol 20:691-695.
McClearn GE, Wilson JR, Meredith W. 1970. The use of isogenic and heterogenic mouse stocks in behavioral research. In: Lindzey G, Thiessen DD,
editors. Contributions to Behavior-genetic Analysis: The Mouse as a Prototype. New York: Appleton-Century-Crofts. p 3-22.
Miller RA, Bookstein F, Van der Meulen J, Engle S, Kim J, Mullins L,
Faulkner J. 1997. Candidate biomarkers of aging: Age-sensitive indices
of immune and muscle function covary in genetically heterogeneous
mice. J Gerontol Biol Sci 52:B39-B47.
Miller RA, Turke P, Chrisp C, Ruger J, Luciano A, Peterson J, Chalmers K,
Gorgas G, Van Cise S. 1994. Age-sensitive T cell phenotypes covary in
genetically heterogeneous mice and predict early death from lymphoma.
J Gerontol Biol Sci 49:B225-B262.
Rose MR. 1991. Evolutionary Biology of Aging. New York: Oxford
University Press.
ACKNOWLEDGMENTS
The author's research on gerontological genetics has been
supported by the MacArthur Foundation Research Network
on Successful Aging and by grants from the National Institute on Aging (AG04948 and AGO9333).
123