INTRODUCTION — Laboratory mice are among the most widely used model systems in biomedical research.
The breeding strategies and genetics of laboratory mice are introduced in this topic review. Detailed information regarding specific mouse strains and breeding constructs are available online at The Jackson Laboratory website.
Other model systems for studying human disease are discussed separately (eg, yeast, worm, fruit fly, zebrafish). (See "Tools for genetics and genomics: Model systems".)
INBRED MICE — Strains of inbred mice are model organisms for genetic studies, with hundreds of inbred strains in existence. For the purposes of genetic investigations, the key properties of inbred mice are that they are essentially isogenic and homozygous. Isogenicity means that sex-matched individuals have the same genotype across the entire genome. Homozygosity means that both chromosomes carry the same allele of each locus.
Concepts of inbreeding were introduced in a classic paper from 1931 [1]. A brief summary of the principles and standards of inbreeding includes the following:
●The generally accepted criterion for a strain to be considered inbred is 20 generations of brother-sister mating, or alternative breeding schemes that lead to an equivalent reduction of heterozygosity. Although historically 20 generations of full-sib inbreeding were considered to be sufficient to fix the genotype of an inbred line, with an expected residual heterozygosity of less than 0.1 percent, it has now been shown that considerably more inbreeding is needed to fix the genome [2].
●Many inbred strains are complex hybrids whose ancestry includes mice that would not have interbred in the wild. For this reason, the amount of genetic variability among laboratory mice is extremely large.
●Nomenclature standards have been developed regarding mouse strain and substrain designations. Substrains are stocks separated after inbred status has been achieved. In general, the strain name is given first, followed by a slash, and then the substrain designation and when applicable, by information regarding mutations or transgenes.
●Not all mice having official names are inbred, and some nominal "substrain" differences are in fact strain differences for which nomenclature exceptions have been made because of historical precedent (eg, C57BL/6 and C57BL/10).
Strictly speaking, new mutations and residual heterozygosity may contribute to non-isogenicity even within an inbred strain, but in most settings this is not an important issue. However, the potential accumulation of differences over time underlies the need to identify both strain and substrain when identifying mice. For these reasons, a single genotype suffices to describe all sex-matched individuals of a single substrain.
F1 animals, the first generation obtained from a cross between parents of different inbred strains, are isogenic but not homozygous. They have all inherited the same alleles from their inbred parents, but one chromosome of each pair is derived from the maternal strain and the other is derived from the paternal strain. The mitochondrial genome is derived from the maternal strain, and it is therefore important to specify the maternal and paternal strains. By convention, in notating crosses, the dam (female) is listed first and the sire (male) second.
These features allow investigation of both linkage and association in experimental mouse crosses:
●Linkage is the determination of the chromosomal location of a gene influencing a trait
●Association is the relationship between a specific allele and the trait
In analyzing the genetic basis of disease, we seek knowledge of both of these complementary properties. A standard mouse intercross is a biallelic study, in which only two alleles (one derived from each parent) are segregating. In contrast, natural human populations are multiallelic.
It is not necessary for an experimental mouse intercross to be limited to two progenitor strains, however. A greater range of alleles can be achieved by increasing the number of progenitors to four or more [3]. In addition, the intercross can be continued beyond F2, increasing the precision with which genes can be mapped [4].
Finally, the dense linkage map of the mouse, together with extensive genotypic data of the inbred mouse strains, makes most mouse crosses fully informative. In contrast, human crosses depend on finding markers in which different alleles can be identified within the study population.
The biallelic nature of many mouse crosses, the ability to control breeding by designating matings, and the known genotypic data of inbred mouse strains greatly simplifies interpretation of the genetic results of experimental breeding relative to the analysis of a human pedigree study. However, this relative simplicity also poses important limitations for the use of inbred mice as models for human genetics.
SPECIAL BREEDING SCHEMES
Recombinant inbred strains — Recombinant inbred (RI) strains are inbred strains established from brother-sister pairs of F2 intercross mice obtained from inbred progenitors [5,6]. The breeding scheme is illustrated in a figure (figure 1) [7,8]. They are isogenic and homozygous, like other inbred strains, and at any given locus they carry either of the progenitor strain alleles.
RI strains are constructed as a series from a single initial intercross. Subsequently, because of crossing over and segregation, the effectively random locations of crossovers during the inbreeding process results in each strain in the series having a different complement of alleles derived from each of the progenitors. Any individual strain in a two-progenitor series is expected to have half of its genome derived from each progenitor strain. More than twenty generations of brother-sister inbreeding are needed for establishment of RI strains. RI strains are powerful mapping tools for monogenic traits and are also useful in mapping polygenic traits [2,9,10].
Congenic strains — Congenic strains are those in which a specific genetic locus is transferred onto a different recipient strain [6,11]. The breeding scheme is illustrated in a figure (figure 2) [7,8]. Since linkage frequently carries neighboring genes along with the target locus into the recipient, this type of strain is generated to evaluate the effect of a gene in isolation from the effect of strain background.
Because the donor segment includes more than just a single gene, only relative isolation of the target gene is achieved. Traditionally, 10 generations of backcrossing to the recipient strain, with selection at each generation for the presence of the desired donor locus, followed by brother-sister mating thereafter, has been the standard of inbreeding for congenic strains. This standard is theoretically equivalent to 20 generations of brother-sister inbreeding.
Many investigators use marker-assisted selection to accelerate the pace at which donor alleles unlinked to the target locus are lost. This approach relies on the fact that while the expected loss of donor alleles is 50 percent at each generation, the actual loss of donor alleles in an individual mouse may be more or less than theoretical expectation. By selecting as breeders those individuals with the least number of donor alleles, the number of generations needed to achieve 99.9 percent homozygosity at unlinked loci can be reduced by three generations, corresponding to a time saving of nearly one year. This approach has therefore been called "speed congenic" breeding [12-14]. Congenic strains first became widely used in immunologic experiments, in which the major histocompatibility complex was transferred between inbred strains. It has subsequently become standard practice to construct congenic strains for quantitative trait loci identified in intercrosses, utilizing negative selection for unlinked donor alleles.
Recombinant congenic strains — Recombinant congenic (RC) strains combine elements of breeding from both RI and congenic breeding schemes. Initially, several generations of backcrossing (usually two) are performed, followed by subsequent brother-sister inbreeding (figure 3) [15-17].
As in congenic strains, the initial backcrosses reduce the contribution of the donor strain to the final genotype of individual RC strains. As in RI strains and in contrast to congenic strains, there is generally no selection for which specific donor chromosome segments are included in RC strains. Conceptually, one might consider RC strains to be "unbalanced" RI strains; while RI strains derive half of their alleles from each progenitor, 2nd backcross (N3) RC strain derive one-eighth of their alleles from the donor progenitor and the remainder from the recipient progenitor. RC strains are used in mapping polygenic traits and studying epistasis, or interactions between genes [10,18-20].
Chromosome substitution strains — Chromosome substitution (CS) strains are intended to facilitate mapping of quantitative traits, by moving chromosomes from a donor strain to a recipient strain. As for RC strains, the contributions of the donor chromosomes are unequal. Unlike the situation for RC strains, the donor contribution is theoretically restricted to a single chromosome, obviating the need for gross mapping of the relevant loci [21]. Constructing CS strains relies on the same principles used in generating congenic strains, except that selection for markers spanning the donor chromosome are used simultaneously, rather than a single marker as in the case of congenic strains. Few CS strains have been bred thus far, so their use has been limited [22-25].
The collaborative cross — The collaborative cross is a notable but ultimately unsuccessful effort to produce a mouse resource for genetic mapping that overcomes the limitations of inbreeding [26]. The objective was to generate a series of 1000 eight-progenitor recombinant inbred strains. The progenitors were chosen to include more than 80 percent of the known interstrain diversity among present inbred mouse strains. F1 animals produced from such strains are isogenic, outbred, and possess haplotype blocks whose length approaches those of natural outbred populations.
Combinational mating among these inbred strains would have allowed 4 to the 1000th power outbred genotypes to be produced in multiple animals, where "4" represents two sexes and two cross directions. The number of potential genotypes would therefore be orders of magnitude larger than is necessary for application of the methods of genomewide association studies (GWAS). (See "Genetic association and GWAS studies: Principles and applications".)
The incipient collaborative cross strains suffered from a high rate of extinction from male infertility, affecting >95 percent of the strains [27]. Approximately half the extinction occurred due to male infertility attributed to introgression of X chromosome regions of either Mus musculus musculus or M. musculus castaneus into backgrounds derived primarily from M. musculus domesticus. Approximately 70 collaborative cross strains are available for distribution, though most retain substantial heterozygosity [28]. Details are available at the UNC Systems Genetics Core website. In spite of the limited number of strains and their residual heterozygosity, they are still useful for some experiments. Their usefulness has been greatest in studying susceptibility to viral infection [29].
OUTBRED MICE — Some commonly used mice are outbred (ie, bred specifically to maintain heterozygosity over much of the genome). Examples include Swiss and CD-1 mice, which are commercially available from major vendors. Such mice are generally healthy, vigorous, and long-lived. In general, such mice are preferred for investigations in which genetic constitution is not an issue. The diversity outcross, a population of mice derived from the collaborative cross, but with subsequent breeding undertaken to preserve heterozygosity, provides a curated outbred population that allows high-resolution genetic mapping [30].
LARGE SCALE MUTAGENESIS — Large scale mutagenesis projects represent an alternative strategy for finding genes that affect specific phenotypes. Typically, single base mutations are produced. Mutagenesis projects depend critically on the ability to recognize abnormal phenotypes reliably and easily, as many mice must be examined to find the desired mutants. Consequently, the mutagenesis strategy relies on generating viable mutants at high frequency, using robust methods to screen mutants, and applying efficient methods for mapping the mutated genes. While random chemical and radiation mutagenesis were common strategies in earlier projects, technical advances have permitted comprehensive, systematic targeted mutagenesis strategies to be undertaken. (See 'Genome editing' below.)
Notable successes have been reported [31,32]. As an example, the Golgi protein GMAP-210, encoded by Trip11, provides a specific highlight of a mouse mutagenesis screen identifying a gene responsible for a human disease. The mutant mouse gene caused shortened limbs, failure of osteogenesis at multiple sites, and apoptosis of growth plate chondrocytes, culminating in lethality [33]. These investigators recognized that the mutant mice resemble humans with achondrogenesis type 1A, and found mutations in the human Trip11 gene in 10 patients with the disorder who were not related to each other.
Various project websites with information regarding other studies include:
●International Mouse Phenotyping Consortium www.mousephenotype.org
●The Jackson Laboratory – www.jax.org
●Centre for Modeling Human Disease at Toronto Centre for Phenogenomics – www.cmhd.ca/enu_mutagenesis/index.html
GENETICALLY ENGINEERED MICE
Technical underpinnings — To understand the various genetically engineered mouse systems, it is necessary to appreciate the technologies used to construct them. Fundamental to all such experiments is the ability to carry out molecular cloning and molecular biological manipulations [34-36]. Specific methods used to introduce engineered DNA into mice are available in standard laboratory references [37,38].
The fundamental technology at the core of most genetically engineered mice is the ability to introduce foreign DNA into pluripotent recipient cells. In practice, this is accomplished either by direct microinjection of DNA into a fertilized egg (picture 1), or by transfection into mouse embryonic stem (ES) cells.
ES cells are pluripotent cells that can give rise to all tissues, including the germline [39-41]. ES cells can be grown in culture and subjected to transfection and selection in the same manner as other cultured cells. Typically, targeted transfection relies upon a combination of selection for resistance to selective agents, Southern blotting, and polymerase chain reaction (PCR) to verify that exogenous DNA has been incorporated into the correct target site. ES cells shown to harbor the desired construct are then injected into mouse blastocysts, to yield chimeric embryos. The blastocysts can then be implanted into foster mothers, which then carry the pregnancy to term.
Transgenic mice — Transgenic mice are those into which foreign DNA has been incorporated into the genome. According to this broad definition, all types of genetically engineered mice are transgenic. In a more narrow sense, transgenic mice have an "extra" gene introduced to accomplish any of several experimental objectives:
●To correct pathology caused by mutation, thus proving that the transgene complements the preexisting mutation [42]
●To introduce a reporter gene under specified genetic control to identify tissues and times at which the included control is active [43-45]
●To introduce an abnormal gene, thus creating a disease model [46]
Simple transgenics, in which no attempt is made to target the construct to a specific site in the genome, can be generated by microinjecting DNA directly into one of the pronuclei of fertilized eggs in vitro. Surviving blastocysts are implanted into foster mothers and pups analyzed for presence of the transgene. Early work was typically performed using outbred eggs, but now much microinjection work is performed using eggs obtained from FVB/N mice, which have large pronuclei and breed well [47].
Knockout mice — In contrast to simple transgenic animals, knockout mice depend upon successful gene targeting to disrupt the target gene [48]. This is achieved by using genetic selection to enrich transfected ES cells for successful targeting. Targeting vectors for mammalian cells use a combination of positive selection (incorporation and retention of a selected marker) and negative selection (loss of a second selected marker) (figure 4). Technical aspects of constructing the necessary targeting vectors have advanced in recent years, and feature bacterial artificial chromosome clones prepared from 129/Sv mice [49] and a "recombineering" host-vector system in which some steps are accomplished in vivo in bacteria [50].
The motivation for doing this is to study the consequences of loss of function of the targeted gene directly in vivo. In contrast to experiments using cultured cells, the knockout allows the consequences of target gene disruption to be evaluated in the context of whole-organism physiology. In this setting, it is possible to study physiological adaptation to the knockout and discover effects in tissues in which pathology might not have been suspected a priori. Comparison and contrast among knockout phenotypes for related target genes help to identify both the unique and redundant functions of their products.
Knock-in mice — Knock-in mice are theoretically similar to knockout mice, with the important difference that an altered rather than a null version of the target gene is substituted for the naturally occurring allele. Knock-in technology allows examination of the effects of different mutations on the same gene. This is particularly informative if the mutations thought to cause human disease result in gain of function rather than loss of function (eg, oncogenes) or if the goal is to investigate a mutant in a single tissue. (See 'Conditional systems' below.)
Genome editing — Notable technical advances have occurred in the ability to create knockout and knock-in animals through the use of zinc finger nucleases, transcription activator-like effector nucleases (TALENs), and the CRISPR/Cas system [51]. (See "Overview of gene therapy, gene editing, and gene silencing", section on 'Gene editing'.)
These tools have greatly reduced the time and expense of creating such animals, and in addition, they can be used in multiple species. They are therefore widely used in zebrafish and rats.
Humanized immunodeficient mice — In spite of their many strengths as a model organism, mice cannot adequately recapitulate all the features of human biology. These limitations are particularly evident in studies of immune function, transplantation, infectious diseases, and tumor biology.
In order to overcome these limitations, investigators have developed the ability to engraft human cells and tissues into mice. These so-called humanized mice rely on exploiting immunodeficient mouse mutants.
Immunodeficient mice have long served as hosts for in vivo study of human tumors and as models for investigation of immune function. The nude mouse (Foxn1nu/nu) was first noted as a spontaneous mutant in 1962, initially recognized for being hairless and having a short lifespan [52,53]. Subsequent work revealed absence of the thymus, impaired T cell development, and ability to tolerate xenografts (tissues from other species) [54,55]. Nude mice have great historical importance, but contemporary humanized mice used to engraft human hematopoietic and immune cells are based on different mouse mutants.
Humanized mice must have the following three principal elements of immunodeficiency to allow robust human immune and hematopoietic cell engraftment:
●A defect in the recombinational machinery necessary for immune cell differentiation, satisfied by loss of function of PRKDC, RAG1, or RAG2.
Loss of function mutations of PRKDC, encoding the catalytic subunit of DNA-dependent protein kinase, result in defective repair of double-stranded DNA breaks, resistance to ionizing radiation, and VDJ rearrangement in B and T cells [56]. Prior work established the severe combined immunodeficiency (SCID, PRKDCscid/scid) phenotype and demonstrated that SCID mice can serve as experimental hosts for engrafting human immune cells [57-60]. This property led to SCID mice rapidly becoming a widely used platform in hematological malignancy, infectious disease, and autoimmunity research, even before the molecular pathogenesis of the SCID mutation was understood. Two additional genes, encoding recombination activating genes 1 and 2 (RAG1 and RAG2), also are needed for VDJ rearrangement, and mice deficient in them show profound impairment of B and T cell maturation, and consequently, immunity [61-64].
●A defect in the shared gamma subunit (encoded by IL2RG) of multiple interleukin receptors.
This defect causes X-linked severe combined immune deficiency (X-SCID) in humans and a similar phenotype in mice [65-68]. Loss of function mutations of IL2RG interfere with high-affinity binding of multiple cytokines including IL-2, IL-4, IL-7, IL-9, IL-15, and IL-21.
●Carriage of a "human-like" allele of SIRPA, encoding signal regulatory protein alpha type 1 [69,70].
This protein is expressed on macrophages and mediates an inhibitory signal that prevents phagocytosis [71]. Non-obese diabetic (NOD) mice coincidentally harbor a polymorphism in SIRPA that resembles the human protein and thereby protects human cells. For this reason, many humanized mice feature a NOD genetic background, although a few strains have a Balb/c background, which also harbors a more favorable SIRPA allele [72]. A few strains have a C57BL/6 background together with SIRPA or CD47 alleles that allow human cell engraftment.
More than 20 mouse strains and stocks capable of being humanized are available from the major commercial laboratory mouse vendors; descriptions can be found on their websites. Additional such strains have been developed by individual laboratories.
To generate humanized mice, an immunodeficient mouse must receive human donor cells. These can be delivered in several ways. Human peripheral blood mononuclear cells or human hematopoietic stem cells can be infused into host mice, which are typically irradiated prior to infusion [58,60]. Alternatively, fetal liver and thymus tissue can be placed under the renal capsule [59].
All of the existing mouse host strains and engraftment strategies display important limitations, and careful planning is necessary to choose the best system to study the relevant biology. Some common problems include development of graft-versus-host disease (GVHD), poor development of lymph nodes, poor maturation of lymphocytes, and limited survival of engrafted tissue. The many available recipient strains represent efforts by numerous investigators to overcome the biological constraints in the context of specific problems encountered in their research.
Conditional systems — Some gene disruptions are lethal or lethal too early in development to allow fruitful investigation of their consequences. A variety of methods have been developed to overcome this class of problems. These are briefly described here.
Modulating transgene expression can be achieved by placing the transgene under the transcriptional control of an inducible promoter. The bacterial tetracycline resistance operon provides such a reagent system [73-75]. The investigator can then regulate transgene expression by titrating administration of tetracycline to the transgenic animals.
A strategy by which knockout expression can be restricted is generation of tissue-specific knockouts. This approach exploits the ability of bacteriophage P1 Cre recombinase to mediate site-specific recombination at specific short sequence elements called loxP sites [76]. Two engineered mice must be produced to carry out such an experiment. One construct introduces the loxP sites into the target gene, flanking sufficient DNA so that its deletion will achieve the desired loss of target gene activity. The second construct, which can be a simple transgene, introduces a functional gene for Cre recombinase driven by a tissue-specific promoter. Mice homozygous for the target gene construct are then mated to mice harboring the tissue-specific Cre construct. Offspring receiving the construct (expected to be 50 percent of the progeny) will express Cre recombinase in a tissue-specific manner, leading to excision of the portion of the target gene flanked by the loxP sites, thus knocking out the gene in that tissue exclusively (figure 5) [77,78]. Many investigators test the biological function of the Cre transgene by mating it to any of several available "reporter" mice [79-85].
Collaborative projects in North America and in Europe have developed conditional knockout alleles for thousands of genes [86-89]. In addition to Cre reporter mice, there are also reporter mice that can identify the anatomic sites at which specific biologic pathways are active. As an example, the TOPGAL mouse expressed beta-galactosidase at sites of canonical Wnt signaling [90], allowing investigators to identify the tissues in which the pathway is active.
The extraembryonic tissues are important in early developmental steps including gastrulation. One group developed a method by which early embryos can be tetraploidized, and chimeric embryos produced from tetraploidized embryos and cultured ES cells [91,92]. Under these conditions, the tetraploid cells give rise exclusively to extraembryonic tissue, while the diploid ES cells produce all of the embryo proper. This approach allows investigators to overcome some early developmental defects, thus allowing later functions of the disrupted gene to be studied [93,94].
Limitations and caveats — Characteristics of inbred strains are highly variable, so knowing the background in which a genetic construct has been studied is essential to interpreting its biology. For technical reasons, many genetically engineered mice have been generated on one of the 129 strains, which recently have been found to be more diverse than previously believed [95-97]. Investigators routinely breed founder animals to a recipient strain, most often C57BL/6. Consequently, the constructs are often studied on a poorly-defined "mixed 129 X C57BL/6 background," without further information regarding the relative contributions of the progenitor genomes, number of generations of subsequent inbreeding, or often even the correct strain information regarding the progenitors. While unfortunate, this is the status of most of the extant literature. Efforts to improve reporting of strain background in the future are underway [98].
For mice in which the construct has not been targeted to a specific locus, incorporation of the transgene may result in insertional mutagenesis, with the resulting phenotype arising not from the transgene, but from disruption of the gene into which the transgene was placed [99]. The transgene's expression may also be variable according to the properties of the insertion site [100].
A third complication is the usual result of multiple copies of the transgene being inserted into the genome, with consequent differences in transgene expression level. These limitations can be addressed by targeting transgenes to specific sites. One site that allows insertion of a single copy of the transgene while allowing transcription to be mediated by sequences included in the targeting vector is the HPRT locus, encoding the salvage purine utilization enzyme hypoxanthine/guanine phosphoribosyl transferase [101,102].
The major limitation of knockouts is that the allele generated is, by design, null. Therefore, while they are useful for establishing the role of the target gene in a pathway, it is not necessarily the case that mutations in the target gene account for human diseases or population variation in downstream phenotypes mediated by it. Knock-in strategies can address this by allowing study of a series of mutant alleles. By virtue of being targeted to the homologous locus, the issue of unintentional insertional mutation does not arise with knockout mice. The limitations related to strain background are significant, particularly because 129 related strains have been the source of most ES cell lines. More recent success in generating successful ES cells from other strains promises to mitigate this problem in the future [103,104].
Another difficulty is that Cre constructs are neither perfectly efficient nor perfectly specific. Thus, "tissue-specific" Cre constructs are sometimes active outside the target tissue, and Cre constructs presumed to be inducible are active in the absence of the inducing substance [105,106]. Furthermore, Cre constructs can exert inherent biological effects, including "Cre toxicity" and Cre-mediated suppression of tumor growth [107-111].
The practical response to these limitations is that when transgenic mice are generated, investigators routinely study animals derived from several different founders. The minimal characterization will generally include estimation of copy number, transgene mRNA level, and transgene protein level. More detailed analysis is then conducted on one or a small number of the transgenic lines. A growing number of studies also use lineage tracing to monitor Cre activity [112]; lineage tracing is emerging as a "best practice" [113]. Appropriate and complete control groups need to be characterized in parallel. Careful investigators will also report the breeding history between founder and the analyzed animals.
SUMMARY
●Advantages of mouse model systems – Laboratory mice are among the most widely used model systems in biomedical research. The key properties of inbred mice relative to genetics studies are that they are isogenic and homozygous. F1 animals, the first generation obtained from a cross between parents of different inbred strains, are isogenic but not homozygous. Isogenicity allows study of multiple animals sharing a single genotype, thus allowing better estimation of the phenotype being studied. (See 'Inbred mice' above.)
●Gene mapping – Recombinant inbred strains, the result of a series of brother-sister matings from an initial intercross, allow gene mapping (figure 1). Congenic strains, in which a specific genetic locus is transferred onto a different recipient strain (figure 2), allow evaluation of a gene in isolation from the effect of strain background. Chromosome substitution (CS) strains facilitate mapping of quantitative traits by moving chromosomes from a donor strain to a recipient strain. (See 'Special breeding schemes' above.)
●Collaborative cross – The collaborative cross preserves the key advantage of isogenicity, while overcoming many of the limitations of inbred mice. This model better approximates the human genetic structure by being outbred and having a preponderance of short haplotype blocks, making this a powerful tool for relating phenotypes to specific DNA variants. (See 'The collaborative cross' above.)
●Transgenic mice and other genetic manipulations – The ability to introduce foreign DNA into pluripotent stem cells is at the core of most genetically engineered mice. Genome editing has greatly reduced the time and expense of creating transgenic animals. (See 'Genetically engineered mice' above.)
•Transgenic mice are those into which foreign DNA has been incorporated into the genome.
•Knockout mice depend upon successful gene transfer targeted to disrupt a specific gene to study the consequences of loss of function of the targeted gene directly in vivo.
•Knock-in technology allows examination of the effects of different mutations on the same gene.