The number one consideration for choosing which genetic technology to use when making a mouse model is the experimental goal itself. In transgenesis, the transgene’s expression is typically superimposed over the endogenous gene’s expression so this approach is traditionally used as a gain-of-function approach. However, transgenesis can also be used as a loss-of-function approach when the expressed cDNA encodes a dominant negative (dn) mutant protein, a partial protein, an anti-sense RNA, microRNA or shRNA. For example, an inhibitory transgenic strategy was effectively used in the heart to investigate the requirement of calcineurin (calcium-activated protein phosphatase that underlies cardiac hypertrophy) in pathologic cardiac growth. In these studies calcineurin function was inhibited through the overexpression of naturally occurring inhibitory proteins Cain/Cabin-1 (the calcineurin inhibitory domain)9
, or RCAN110, 11
or through the overexpression of a dn-calcineurin12
. The phenotypes observed in these calcineurin inhibitory transgenic lines, which showed inhibition of pathologic cardiac hypertrophy, closely matched the antihypertrophic phenotype of calcineurin Aβ (Ppp3cb
) gene-deleted mice, which were designed later. This comparison suggests that the transgenic approach, despite involving gross overexpression, was effective at uncovering the phenotype of calcineurin loss-of-function in the heart13
. Indeed, the more traditional usage of cardiac transgenesis, in which an activated mutant form of calcineurin was overexpressed in the heart with the α-MHC promoter, showed massive cardiac hypertrophy14
Gene targeting is most often used to create deleted alleles for loss-of-function experimental strategies; however, KI models can be made in which a mutant version or selection cassette replaces the endogenous gene that in some instances can produce a gain-of-function like effect15
. Replacement by gene targeting also employs the endogenous regulatory elements of the targeted loci, so that if a KI mutation is used, expression occurs at physiologic levels. For example, a gene-targeted KI strategy was used to create a mouse model of inherited hypertrophic cardiomyopathy (HCM) by replacing a single amino acid in the endogenous α–MHC gene locus, R403Q16
. This mouse was instrumental in better understanding the molecular-genetic basis for the HCM clinical phenotype17–19
. By comparison, Leinwand and colleagues generated a transgenic mouse model of HCM by overexpressing a mutant αMHC protein in the heart using a rat α-MHC promoter20
, which revealed a disease phenotype consistent with the R403Q mutant KI mouse16, 21
. These results again suggest that transgenesis can be just as informative as gene-targeting if properly designed, and typically many times faster with considerably less cost. Despite these considerations, the gene-targeting based approach (such as a KI mutation) is often more elegant and can be more physiologic.
Gene targeting technology can also be used to “knock-in” a reporter gene (i.e. β-galactosidase) into a locus of interest to identify temporal and spatial patterns of gene expression during cardiac development or in response to stress. For instance, to study the role and expression of islet1 (Isl1) in murine heart development a nuclear β-galactosidase construct was “knocked-into” the isl1 locus22
. This mouse was then used to visualize myocyte lineage subdomains that express β-galactosidase as a surrogate for Isl1 during development.
While both transgenesis and gene-targeting are powerful experimental approaches, there are pitfalls associated with each that should factor into an investigator’s decision-making. These pitfalls are listed in and more thoroughly reviewed below (). Perhaps the main drawback of choosing gene targeting is the considerable amount of time and money it takes to generate the mouse, assuming that one will be successful in targeting the desired locus in the first place (not a guarantee). Another issue with gene targeting is that many genes play vital roles during embryonic or early postnatal development, so that traditional somatic targeting in the mouse (absent from zygote stage onwards) might produce lethality, precluding an analysis of an adult heart disease-related phenotype. Indeed, we were unable to ascertain the adult cardiac hypertrophy-associated phenotype of ERK1/2 in the heart using traditional gene-targeting because Erk2
null mice die early in embryonic development23
. We later bypassed this developmental lethality issue using Cre-LoxP technology to generate heart-specific deletion of Erk2
in the Erk1−/−
. In addition to embryonic lethality, another problem is that the adult cardiac phenotype observed in a traditionally targeted KO mouse could be influenced by loss of gene function in other organs or tissues. Finally, replacement of the endogenous gene at a given locus can alter transcriptional patterns of neighboring genes due to the deletion of their critical regulatory elements even with a precise targeting construct or from interference with the neomycin cassette and its promoter. Knowing as much as possible about the genomic locus of interest through comparative genomic programs can guard against this situation and is further discussed in the supplement (Supplemental Information
). A classic example of this was identified when three different labs independently deleted the muscle regulatory factor MRF4. Interestingly, 2 of the 3 MRF4 KO models were lethal during development while the other was not. In the 2 non-viable models, expression from the neighboring Myf5
gene was also absent despite the Myf5
locus being intact25–27
Experimental considerations for genetically modified mice.
Transgenesis also has various technical challenges due to the fact that the transgene randomly inserts into the genome and that expression levels are typically higher and can be considered “non-physiologic” (). However, we have generated transgenic lines where expression almost exactly approximates the known increase in expression of a disease-associated gene in the heart. For example, thrombospondin 4 is induced in the heart following many different disease-inducing stimuli, and this level of induction was approximated by transgenesis with the α–MHC promoter, accurately revealing its disease modifying phenotype28
Another issue that can seriously compromise a genetic based experiment in the mouse is if more than one gene or transgene to be simultaneously manipulated resides on the same chromosome. The easiest way to understand the issue is by example from our own laboratory. We were attempting to develop double null mice between the δ-sarcoglycan (Sgcd) and the JNK2 (Mapk9) deleted loci to determine if loss of JNK2 protein might impact muscular dystrophy pathogenesis in mice lacking δ-sarcoglycan. We crossed the mice, but their offspring were never genotyped as double homozygous. Only double heterozygotes were ever identified. Checking the genomic database we easily surmised that the reason was because both genes reside on mouse chromosome 11. This same issue can befall a transgene if one is trying to cross it into a homozygous null gene-deleted background. Moreover, transgenes are rarely mapped for chromosomal insertion, and when one fails to identify homozygous nulls with the experimental transgene of interest, most of us would interpret it as a lethal event due to the biology associated with the cross, when actually it is because transgene and targeted allele are on the same chromosome. However, if the genes/transgenes of interest are far enough apart on the same chromosome, meiotic cross-overs can occur with enough breeding, which will allow for success in obtaining double nulls or a null loci with a hemizygous transgene. Finally, if maintained in the hemizygous state, 2 transgenes are easily crossed even if they reside on the same chromosome, but neither can be made homozygous without obtaining cross-overs (and if the loci of interest are too close together on one of the chromosomes, such events can be extremely rare).