Crusio, Wim, E (1996) Gene targeting studies: New Methods,
old problems. Trends in Neurosciences
Commentary by Wim E. Crusio
Génétique, Neurogénétique et Comportement
URA 1294 CNRS, Université de Paris V
45 rue des Saints-Pères
75270 Paris Cedex 06
gene targeting, neurobehavioral genetics, hippocampus, inbred
mouse strains, neurological mutants, acetylcholine
Send correspondence to: Dr. Wim E. Crusio at the above address.
Tel. 19 33 1 42 86 22 06; Fax. 19 33 1 42 86 22 50; e-mail: firstname.lastname@example.org
Techniques to create transgenic organisms
or animals with targeted mutations ("knock-out" mutants)
have become increasingly important tools in the neurosciences
over the last few years. As always, new techniques, besides providing
new tools to investigate problems or to test hypotheses, also
give rise to unforeseen difficulties and it takes some time for
researchers to become aware of this. Gene targeting techniques
provide no exception and Gerlai's1 alert is very timely indeed.
The complications described by Gerlai are, not very surprisingly,
strongly reminiscent of those encountered in the study of spontaneous
mutations in the mouse. In that field congenic strains have been
used since many years2. Congenic strains are obtained by repeatedly
backcrossing a mutant to an inbred strain. If we now consider
the possible complications that may be encountered with this approach
we may, in fact, distinguish two rather different types of "background"
effects. As I will show in the following, the distinction between
the two is crucial.
Hitchhiking donor genes
As noted by Gerlai, some residual donor genes that are closely linked to the mutated gene may still be present in a congenic line even after numerous generations of backcrossing. Of course, the presence of these contaminating hitchhiking genes may bias our experiments in a completely unpredictable direction, possibly leading to false positive or negative results. Up till now, no real solution to this problem was available, apart from continuing the backcrossing procedure over as many generations as feasible. Using modern techniques of gene transfer, Gerlai proposes several elegant possibilities by which researchers might control for confounding effects due to these kind of genes. I would like to suggest here an additional one.
In a way, targeted mutations are analogous
to another widely used technique in the neurosciences, namely
brain lesions. Knock-out experiments have in common with lesion
experiments the assumption that the function of the lesioned structure
equals the dysfunction of the residual organism. Also, in both
types of studies we correlate certain effects with the presence
or absence of an impaired structure (i.e., brain structure or
gene). Obviously, correlational studies gain much in power if
more than just two data points are available. For instance, we
might exploit naturally occurring, non-pathological variations
in neuroanatomy between individuals to uncover brain-behaviour
relationships3. This approach has been named microphrenology by
Lipp4 and may be augmented profitably by adopting a genetic strategy5.
An analogous way to increase the number of data points in gene
targeting studies might be the addition of transgenic animals.
We could then compare the effects of underexpression (null mutants),
normal expression (wild type), and overexpression (transgenics)
of a certain gene. If different transgenic lines with different
numbers of copies of the transgene are available, the number of
data points can be increased even further. If this approach would
render consistent results, we might be reasonably certain that
the observed effects were due to the manipulated gene and not
to genetic contamination by residual donor genes.
The recipient background
Second, there are the possible effects of the genetic background in its more classical sense of the recipient genotype (not necessarily inbred) to which the mutation has been introduced. In essence, we are here dealing with an interaction: the phenotypical effect of the mutation depends on the genetic background. A recent example of such a case was provided by the EGF knockout mice6,7, that showed very different phenotypes depending on the strain background upon which the EGF null mutant had been transferred. The latter result apparently surprised many, which in itself is quite amazing and telling because this should not have been necessary. Genotype-treatment interactions are a well-known phenomenon in the field of neurobehavioral genetics, be they differential expression of mutant genes depending on the genetic background8-10, or unequal effects of brain lesions11,12 or divergent effects of pharmacological treatments13 in different inbred strains.
Many researchers will probably consider such interactive effects a nuisance and, in consequence, choose to work with a genetically heterogeneous population. For a number of reasons, this would be a pity. First of all, although no interactive effects will be observed in such a population, this would be for the obvious reason that such interactions simply cannot be detected by such an experimental design. Of course, it is often argued that a heterogeneous population is more representative of the human population that it is supposed to model. What is overlooked is that such is only true if we are interested in human beings as a population, but not as individuals. If I were to suffer unpleasant secondary effects from some drug prescribed to me by my physician, it would be small consolation indeed to know that in the mean, this particular drugs has beneficial effects. To ignore an important experimental factor, in this case by using a genetically undefined population of experimental subjects, clearly is not an optimal research strategy.
Second, genotype-treatment interactions may provide the neuroscientist with a welcome additional tool. An instructive example has been provided by van Abeelen, who used a pharmacogenetic approach to investigate the role of acetylcholinergic (ACh) neurotransmission in the hippocampus in the regulation of mouse exploratory behaviour14,15. He used the inbred mouse strains C57BL/6 and DBA/2 that differ consistently in their levels of several exploratory acts in a novel environment: the former rates high, the latter low. Intrahippocampal injections with the anticholinergic drug methylscopolamine depressed the scores of the C57BL/6 strain, whereas those of the DBA/2 strain were enhanced. Equal treatments yielded opposite effects. In contrast, similar treatment with the acetylcholinesterase (AChE) inhibitor neostigmine depressed scores in both strains. Van Abeelen15 concluded that there exists a genotype-dependent cholinergic mechanism in the hippocampus that controls exploratory behaviour in mice. A functionally well-balanced ACh/AChE ratio appears to promote high exploration scores in C57BL/6 animals. Any injection with drugs that cause an imbalance in this ratio in either direction thus leads to a decline in exploratory activity. In DBA/2 mice a disequilibrium of this ratio (excess of ACh) leading to low levels of exploration was postulated. Correcting the imbalance by injecting anticholinergics then results in augmented exploration. Additional support for van Abeelen's hypothesis of hippocampal malfunction in DBA/2 mice has recently been provided16.
From the preceding, we may even hypothesize
a probable differential expression of an ACh receptor null mutation:
it might be expected to depress exploratory behaviour when expressed
on a C57BL/6 background but to augment it on a DBA/2 background.
If we should use, for instance, an F2 generation between these
two strains as genetic background for our null mutation, then
any conceivable result might be obtained, depending on the exact
composition of our particular sample.
As the above examples show, the study of genotype-treatment
interactions and of naturally occurring interindividual variability
may greatly enhance our understanding of the functioning of brain
systems. A final note of caution may therefore be at his place
here. Modern techniques like gene targeting and transgenesis provide
exciting new opportunities for neuroscience research. However,
in our excitement we should not forget that some questions might
sometimes just as well, if not better, be addressed using less
flashy and less fashionable techniques.
The preparation of this article was supported by the CNRS (URA 1294) and DRED (Université Paris V René Descartes).