Genetic quality assurance and genetic monitoring of laboratory mice and rats: FELASA Working Group Report

Genetic quality assurance (QA), including genetic monitoring (GeMo) of inbred strains and background characterization (BC) of genetically altered (GA) animal models, should be an essential component of any QA programme in laboratory animal facilities. Genetic quality control is as important for ensuring the validity of the animal model as health and microbiology monitoring are. It should be required that studies using laboratory rodents, mainly mice and rats, utilize genetically defined animals. This paper, presented by the FELASA Working Group on Genetic Quality Assurance and Genetic Monitoring of Laboratory Murines, describes the objectives of and available methods for genetic QA programmes in rodent facilities. The main goals of any genetic QA programme are: (a) to verify the authenticity and uniformity of inbred stains and substrains, thus ensuring a genetically reliable colony maintenance; (b) to detect possible genetic contamination; and (c) to precisely describe the genetic composition of GA lines. While this publication focuses mainly on mouse and rat genetic QA, the principles will apply to other rodent species some of which are briefly mentioned within the context of inbred and outbred stocks.

While this publication focuses mainly on mouse and rat genetic QA, the principles will apply to other rodent species some of which are briefly mentioned within the context of inbred and outbred stocks.

Standardized laboratory rodents 1.1 Inbred strains
The International Committee on Standardized Genetic Nomenclature for Mice and The Rat Genome Nomenclature Committee considers a strain inbred 'if it has been propagated by systematically mating brothers to sisters (or younger parent to offspring) for 20 or more consecutive generations, and individuals of the strain can be traced to a single ancestral pair at the twentieth or subsequent generation'. At this point, animals within the population will average ≤ 2% residual heterozygosity, and the individuals may be regarded as genetically identical (isogenic) 1 . However, it has been estimated that 24 generations of sib-mating are needed to reach a heterozygosity rate <1% and 36 generations to reach (almost) complete isogeneity 2 .
Most inbred mouse and rat strains commonly used in research have gone through tens of generations of inbreeding. Some strains have been bred in this manner since the beginning of the last century, meaning for over 200 generations (for example, in 2018 DBA/2J reached F224).
Isogeneity implies histocompatibility, meaning the strains are syngeneic. Syngeneic animals will permanently accept tissue transplantations from any individual of the same strain and sex. Unlike cloned animals and monozygotic twins (which are 100% identical for all genomic loci), inbred rodents, besides being isogenic, are also homozygous at almost all genomic loci (polymorphic in the founder ancestors). This is because after a few tens of generations, one allele segregating at a given locus becomes fixed, whereas the others are lost. Fixation occurs when one allele, present at generation F, is absent in at least one of the two breeders mated to produce generation F + 1 causing its permanent loss. Which alleles become fixed or lost usually depends on chance. Overall, each inbred strain represents a unique, although fortuitous, assortment of alleles 3 . If a strain were to be remade from scratch, using the same founders, after the same 20 generations of inbreeding it would create a genetically distinct strain due to the random assortment and fixation of alleles.
The most striking consequence of fixation of alleles in inbred mice is the diverse coat colours of distinct strains. More important, however, are induced physiological changes, which can either benefit specific research applications or confound and negate experimental results. Each inbred strain has a unique collection of characteristics that should be carefully considered when selecting an animal model. For example, homozygosity at particular alleles renders some strains blind (e.g., the Pde6b rd1 mutation) or causes age related hearing loss (e.g., the Cdh23 ahl mutation).
Some strains are susceptible to spontaneous or induced tumour development, whereas others are resistant to tumour formation. Some strains are aggressive and others are relatively tame, and so on. Baseline phenotypic data for the most common Project-Rat (NBRP-Rat, Japan). Comprehensive information about the genetics and the biology of the most common strains, describing their specific genotypes and phenotypes, are available online 11 12 13 14 15 16 (websites listed in Table 2). In addition, MouseMine and RatMine are continuously updated data warehouses that encompass a variety of source databases making integrated data connecting genes to phenotypes more readily accessible.

Outbred stocks
Outbred stocks are populations of laboratory animals that differ from inbred strains in that they are genetically heterogeneous. According to the standard definition, outbred stocks are 'closed populations (for at least four generations) of genetically variable animals that are bred to maintain maximum heterozygosity'. Compared with inbred strains or F1 hybrids, the genetic constitution of a given animal, taken randomly from an outbred stock, is not known a priori. However, all of the animals in the group share characteristics (identity), for example being albino (although not all the stocks are albino), good breeders, and relatively tame; features that make these animals very popular as foster mothers for assisted reproductive techniques.
Examples of outbred stocks of mice are ICR (CD-1), CFW, and NMRI (all derived from the original 'Swiss' mice imported to the USA by Clara J. Lynch in 1926) and (non-Swiss) CF-1. Examples of outbred rat stocks are Sprague Dawley (SD), Wistar (WI), and Long-Evans (LE). In contrast to inbred strains, which are usually well characterized and described in databases of research centres, there is no comparable source of detailed information (for example, allele frequencies of specific selected markers) for outbred mice or rats. Since outbred stocks are not genetically defined, quality control is commonly based on assessing expected phenotypic traits, such as coat colour, growth, and reproductive characteristics, based on data from the large colonies of commercial breeders.
Outbred animals are typically bred to maintain a defined level of population heterozygosity and to avoid inbreeding 17 . Several breeding schemes for rigorous outbreeding have been developed and should be applied instead of experimenting with random mating. Frequently used systems include the so called "rotational breeding" 18 19 . A key element of these schemes is dividing the colony into a fixed number of equally sized groups, determined by which females and males will be mated to each other. A constant number of progeny per parent of the breeding stock is selected to avoid unintended directional selection: the greater the number of breeders in a stock, the smaller the variations in allele frequencies (genetic drift) at each generation 18  Outbred stocks of other laboratory rodents are also available, including guinea pig, Syrian hamster, Chinese hamster (Cricetulus griseus), gerbil, cotton rat (Sigmodon hispidus), and sand rat (Psammomys obesus).

F1 Hybrids
F1 hybrids result from the outcross of two separate inbred strains and are heterozygous at all loci for which the parental strains harbour different alleles. F1 littermates are genetically identical (isogenic) and are histocompatible. They will permanently accept tissue transplantations from both parental strains, from their littermates, and from their offspring; however, the parental strains will not accept a graft from the F1 hybrids. F1 hybrids exhibit hybrid vigour, making them an attractive choice in some protocols, for example, DNA pronuclear microinjection (e.g., B6D2F1 mice). However, when they are intercrossed, the resulting F2 generation is genetically heterogeneous due to segregation of polymorphic loci.

Other standardized strains of mice and rats
In the last few decades, there has been a dramatic increase in the use of congenic strains, particularly for the maintenance of transgenes and mutant genes, including knockouts (KO), knockins (KI), and spontaneous mutations. Congenic strains are produced by crossing two strains: the donor strain that carries the allele or chromosomal region of interest, and the recipient or background strain that will receive the locus of interest. F1 offspring generated by crossing donors and recipients are then backcrossed to the recipient strain. Offspring that carry the allele of interest are identified and again crossed to the background strain. This process is typically repeated for 10 or more successive generations (Figure 1), unless markerassisted backcrosses (speed congenics) are used. Repeated backcrosses result in the chromosomes of the background strain progressively replacing those of the donor strain, except for a chromosomal region that carries the allele of interest. For this particular chromosome, the chromosomal segment containing the allele of interest is reduced in size only when a recombination event occurs that replaces a piece of chromosome of the donor strain with the homologous segment of the background strain. Consequently the chromosomal segments flanking the selected locus tend to remain associated with it and this is a limitation of the congenic lines due to the potential presence of modifier genes in this segments, the so-called "flanking gene problem" 26 .
Notably, when a mutation occurs in the breeding nucleus of an inbred strain, the new strain differs from the original only by that specific mutation. These two strains are said to be co-isogenic. Co-isogenic strains are extremely useful for gene annotation because they allow a comparison of the phenotypes of two allelic forms of a particular gene without influence from the genetic background. The albino C57BL/6-Tyr c strain is a co-isogenic strain that is popular for making easily recognizable, chimeric mice from C57BL/6 ES cells injected into albino C57BL/6-Tyr c /Tyr c blastocysts 27 .
There are several other types of strains, used almost exclusively by geneticists. For example, consomic strains, also designated chromosome substitution strains (CSS), are a variation of the congenic strains concept. Here, the introgressed DNA is a complete chromosome, rather than a piece of chromosome flanking a given gene 28 .
Recombinant inbred strains (RIS) are used mainly for gene mapping and are developed by crossing two parental inbred strains to generate F1 hybrids and then by intercrossing the F1s to generate F2s. Finally, randomly chosen F2 animals are sib mated for twenty or more generations to develop a group of related inbred strains 29 .
RIS derived from the same parental strains are grouped into sets. For example, the C57BL/6 × DBA/2 (BXD) is currently the largest mouse RI set and has ~90 strains.
The HXB and BXH rat sets (derived from SHR and BN-Lx strains) are also available for genetic studies. The Collaborative Cross is a variation on the RIS concept and is being established by crossing eight founder strains, thus providing a much higher power of resolution and level of genetic diversity than current RIS sets. The Collaborative Cross will provide a new population model designed for analysing complex traits and diseases by covering 90% of the known genetic variation in laboratory mice (The Complex Trait Consortium) 30 31 .

Genetically altered (GA) rodents
Before presenting the different types of GA rodents, it is worth to mention that there are basically two different approaches for characterizing gene function. Forward genetics (from phenotype to genotype) aims to characterize the gene alteration that is responsible for a specific mutant phenotype (typically from spontaneous or chemically-induced mutations). Reverse genetics is the opposite approach and aims to characterize the function of a gene by analysing the consequences (at the phenotypic level) of alterations normally engineered by researchers at the DNA level.
This section introduces the four basic types of GA rodents, those created by: (i) pronuclear microinjection, (ii) vector-mediated transgenesis (iii) homologous recombination in ES cells, (iv) gene editing nucleases, and (v) either chemically induced or spontaneous mutations. Detailed descriptions of the technologies used to create GAs have been published 32 33 34 35 . Please also see the information provided  Table 2). If a model is not available, then the most optimal method for generating the GA rodent must be selected.

Transgenesis by pronuclear microinjection
Transgenic mice were introduced in the early 1980s 36 37 38 , and were the first transgenic animals. It is advisable to use the term 'transgenic' only for animals whose genomes have been altered by the random insertion of DNA 1 . Transgenic rodents are almost exclusively created by the pronuclear microinjection of foreign DNA fragments directly into one of the two pronuclei of one-cell embryos (zygote), a technique that is still widely used. In this process of additive transgenesis, the microinjected transgene randomly integrates into the genome as a single copy or as a concatemer with variable copy number. The mouse and rat models created with this system typically express or in the resultant concatemer, overexpress a transgene placed under the control of a tissue-specific, developmental-stage-specific, or ubiquitous promoter (along with other regulatory elements), all contained in the transgene DNA construct.  39 . In order to achieve a pure genetic background (recommended), the transgene must be introduced into embryos derived from an inbred strain, such as FVB/N, which is widely used because its zygotes possess large and prominent male pronuclei, and the females are excellent breeders that produce large litters 40 .
However, care should be taken when selecting the background strain and its intended research area. If the background utilised during production is inappropriate, a costly (time and animal numbers) process of backcrossing may be required (e. g., FVB/N is not appropriate for many behavioural studies).   When the loxP sites are in the same orientation and on the same strand (in cis), the intervening stretch of DNA is excised. When two loxP sites are in the opposite orientation and on the same chromosome the intervening DNA segment is inverted.

Targeted mutagenesis by homologous recombination using ES cells
Finally, when the loxP sites are on two different chromosomes (in trans) the recombinase generates a reciprocal translocation. Other strategies for creating conditional mutants include the Flp-FRT and the Dre-Rox systems 49 50 .
The Cre transgene can be made inducible, adding more sophistication to the system.
The tamoxifen-inducible Cre ERT2 , which can be activated in a spatio-temporal manner by administration of tamoxifen, is widely used 51

Gene editing using nucleases
Over the last ten years, a number of new techniques have been developed for the production of targeted mutations using engineered nucleases. These techniques, briefly described below, provide ES cell-independent methods to create targeted mutations in laboratory mice, rats, and other species 54 .

Zinc-finger nucleases and transcription activator-like effector nucleases
To make mutations using zinc-finger nucleases (ZFN), two complementary and sequence-specific multi-finger peptides containing the FokI nuclease domain must be designed. Each peptide is designed to recognize a specific DNA sequence spanning 9-18 base pairs (bp) on either side of a 5-6 bp sequence, which defines the targeted region. When injected into a pronucleus or cytoplasm of zygotes, the ZFN assemblies bind tightly, one on each strand, on both sides of the target site.

The CRISPR/Cas System
The CRISPR (clusters of regularly interspaced short palindromic repeats) -Cas system, commonly implemented as CRISPR/Cas9, is based on a primitive defence mechanism that allows bacteria and archaea to fight against infection from viruses, plasmids and phages 61 62 . CRISPR-based guide RNAs (gRNAs) are designed to target a Cas endonuclease to cut DNA at the desired site through RNA-guided DNA cleavage.
The RNA-guided endonucleases can be engineered to cleave virtually any DNA sequence by appropriately designing the gRNA, for example to generate KO mice 63 .
CRISPR/Cas technology has several advantages over ZFNs and TALENs. The main advantage is the ease of design and the flexibility of using a sequence-specific RNA interacting with the Cas enzyme instead of a complex sequence-specific protein (DNA-binding domain) fused to a nuclease. Also, mutations in multiple genes can be generated in a single step by injecting mice with multiple gRNAs that simultaneously target different genes 64 . Such multiplex gene editing has been successful in cells, as well as mouse and rat embryos 63 65 66 67 . CRISPR/Cas9 has been used to create insertions, deletions, and point mutations. The system is highly flexible, fast, and efficient, and is revolutionizing genomic engineering in mammals 68 74 or injected into either the cytoplasm or pronuclei of 1-cell, or 2-cell stage embryos 75 , thus avoiding the use of ES cells and chimeras. However, as each engineered animal is unique, this technology requires extensive sequence analysis to characterize multiple putative founders to ensure the presence of the desired mutation and the absence of undesired on-and off-target mutations or unpredictable larger genome alterations 76 77 78 , while also identifying mosaic founders (G0). Once identified, the selected founder should be bred with wild-type animals to evaluate transmission of the mutation to their offspring. The discovery of the extraordinary virtues of the alkylating agent N-ethyl-N-nitroso urea (ENU) as a mutagen was a milestone in the history of mouse genetics.

Spontaneous and chemically induced mutations
Researchers using ENU have generated and propagated numerous mutant alleles for protein-coding genes, thus establishing a precious tool for genome annotation.
Because ENU typically creates point mutations, it has been widely used in forward genetic screens. The major drawback of ENU-induced mutagenesis is that it creates random mutations rather than targeted mutations. Several projects have been undertaken to systematically and extensively phenotype the offspring of ENUmutagenized males. Large ENU mutagenesis programmes have been conducted in Germany, England, and the USA 79 .

What to ensure after (in-house) generation or upon arrival?
The possibility of crossing different GA lines combined with the increasing complexity of targeting approaches has greatly increased the number of available GA models. The need to cross different GA lines together for a particular study generates additional complexity, especially at the genetic background level. Many mutants have been and are still generated on a hybrid genetic background. Therefore, it is essential to keep adequate records of detailed information for all genetically modified strains.
This information must be transferred with the strain to all collaborators and users.
The most important information includes the correct strain name, a complete description of the mutation, the background of the animals, a genotyping protocol, and observed phenotypic changes. Together, these provide the minimum information for the recommended 'rodent-passport', and several forms have been designed for cataloguing this information. We recommend the data sheet developed by the FELASA Working group on the refinement of methods for genotyping geneticallymodified rodents 80 .
Every mutant strain name must provide precise information on the affected gene, the should also be provided electronically to facilitate analysis of the modified locus. For published strains, a reference to the publication originally describing the mouse model must be included.

Quality assurance for genotyping samples
Genotyping GA mice is a routine procedure. The FELASA Working group on the refinement of methods for genotyping genetically-modified rodents has created a detailed report on current methodologies for sample collection and genotyping 80 .
Some important points are highlighted below.
Rodent genotyping protocols should be specific, simple, easy, and robust, and avoid animal harm. The method of sample collection and genotyping must not stress the animals. Genotyping is the only reliable way to maintain a colony over time and to share bona fide mouse strains between scientific collaborators. One important prerequisite is that a reliable animal identification system is used, as no genotype information is valid without an unequivocal way to link it to the individual animal at all times. Mice and rats must be marked with a well-defined identification code that,

Origin and consequences of genetic variation
A serious challenge facing rodent animal facilities is keeping inbred strains genetically pure and GA lines on a defined background. Changes in the genetic constitution of inbred strains can be produced by (i) contamination by an accidental outcross or multiple outcrosses, and (ii) genetic drift due to residual heterozygosity or fixation of de novo spontaneous mutations.

Genetic contamination
The

2 Spontaneous mutations and polymorphisms
Spontaneous mutations are a source of uncontrolled genetic variation that is often impossible to detect by simple phenotypic observation or routine GeMo. Estimated rates of spontaneous mutation in mammals range from 10 −5 to 10 −8 per locus per gamete, depending on whether being assessed based on breeding or sequencing data.
If the mutation rate is 0.5-3 x 10 −8 per bp per generation, then approximately one protein-coding mutation per generation is expected to arise through genetic drift 87 .
Genetic polymorphism is the presence of alternative DNA sequences (alleles) at a locus among individuals, groups, or populations, at a frequency > 1%. In the laboratory rat and mouse, analysis of these genetic variations has evolved with the need for genetic markers in linkage studies. An enormous number of genetic markers, polymorphic between inbred strains, has been instrumental for developing genetic maps and identifying genes by positional cloning, but these markers can also be used for GeMo and background characterization 1 . Two types of genetic markers are commonly used in association studies and genetic quality control: microsatellites and single nucleotide polymorphisms (SNPs) (see Section 5.1).

Genetic drift and the generation of substrains
While permanent inbreeding effectively eliminates a proportion of new mutant alleles, another undetected fraction may become progressively fixed in the homozygous state, replacing the original allele, a process known as genetic drift.
Genetic drift contributes inexorably to strain divergence and the generation of substrains when the same strain is propagated independently in different places 88 .
Examples of mouse substrains are abundant, for example there are ~10 documented BALB/c substrains and ~15 C57BL/6 substrains including the J and N substrains from The Jackson Laboratory (Jax) and the National Institutes of Health (NIH), respectively 89  Each lab code contains one to five letters and identifies the institute, laboratory, or investigator that produced and/or maintains a particular strain.

Undesirable passenger mutations
Mutations that are hidden in the genomes of substrains or GA lines and can affect the outcome of an experiment are sometimes referred to as passenger mutations 93 . There

Importance of using standard nomenclature
Naming and describing inbred strains with standard nomenclature is critically important 81 104

Genetic quality control programmes
The current gold standard for genetic quality control of laboratory rodents depends on polymorphic genetic markers to distinguish between different genetic backgrounds. Genetic markers are specific DNA sequences with a known location on a chromosome, and are essential tools for genetic quality control. Genetic quality control is essential to determine the genetic composition of an animal and to screen for uniformity and authenticity of a strain 3 , to detect genetic drift, and genetic contamination and to monitor the progress of breeding programmes and to select future breeders.

Marker Systems
Many polymorphisms have been described in the mouse and rat; however, only microsatellites and SNPs are used as genetic markers in current QA programs.
Historically, biochemical markers, especially enzyme polymorphisms, were used extensively for GeMo of inbred strains. Immunological markers, especially those of the Major Histocompatibility Complex (MHC), called the H2 complex in mice and the RT1 complex in rats, were also used to verify genetic authenticity. The advent of DNA profiling/fingerprinting methods introduced restriction fragment length polymorphism (RFLP) analysis using probes directed to minisatellite DNA sequences.
RFLP technology was quickly adapted for GeMo of inbred strains 105 , but has been largely replaced with less cumbersome, modern methods. It is still early to assess whether whole-exome or whole-genome sequencing will be useful for general QA purposes. Currently, whole-exome sequencing provides a robust method for discovering hereditary factors controlling rare Mendelian disorders in humans, as well as new mutations previously mapped through positional cloning in mice and rats 87 , and could be useful for the characterization of substrains.

Microsatellite markers
Microsatellite markers, also known as Simple Sequence Length Polymorphisms (SSLPs) or Short Tandem Repeats (STRs), are still used in modern GeMo programs because they are inexpensive and easy to type 106

Single Nucleotide Polymorphisms (SNPs)
SNP genotyping is an alternative to microsatellites that is now widely used for GeMo.   119 .

Genetic monitoring (GeMo) of inbred strains and outbred stocks
It is important to observe animal phenotypes when monitoring breeding colonies.
Most phenotypic traits are polygenic, resulting from the effects of multiple genes, some of which are pleiotropic and influence more than one trait. The phenotype can also be modified by environmental factors, such as the conditions in which the animal was born and raised. Therefore, a robust monitoring plan of quantitative traits should be based on measurements of an appropriate number of animals.
Monitoring should include observation of spontaneous (innate) behaviour, which differs between inbred strains of laboratory rodents; maternal care and rearing behaviour, which can be assessed during routine breeding work; and obvious phenotypes such as coat colour should always be considered. Detailed information on the genetics of rodent pigment variation have been summarized in textbooks 120, 121 and genome databases ( Table 2). Examples in rat inbred strains are Type 2 diabetes in the Zucker diabetic fatty rat (ZDF) and hypertension in the spontaneously hypertensive rat (SHR) 85 .

Robust GeMo programs also monitor reproduction parameters including
For outbred stocks, GeMo helps preserve the genetic heterogeneity and allele pool of a colony. This complex process requires analysing a large number of animals and comparing this data with historical data documenting the alleles present, their frequency, and the level of heterozygosity in that particular colony. In some cases, the results can reveal a loss of genetic variability resulting in a colony with very low heterogeneity. The degree of genetic heterogeneity in outbred colonies depends on their history. Low heterogeneity can result from poor selection of future breeding stock, deviation from approved (rotational) breeding systems or the bottleneck effect caused by a small breeding pool, as is common when a small group of breeders is imported or being used to rederive a colony. In contrast, high heterogeneity can result from a recent outcross. In general, outbred stocks are characterized by measuring phenotypic traits and calculating the corresponding mean and standard deviations. Essentially, genetic control of outbred stocks is directed at avoiding inbreeding and stabilizing genetic diversity over many generations.

GeMo of inbred mice and rats bred in-house.
The best recommendation here is to purchase animals from reliable vendors and replace the breeding stock with animals from the same vendor after 10 generations, rather than to maintain independent colonies of classical inbred strains. As an additional benefit, using animals from the same vendor prevents the formation of substrains harbouring potential mutations. Nevertheless, in-house colonies should always be tested with a small set of informative microsatellite markers or SNPs to confirm integrity.

Using a small panel of microsatellites (SSLPs)
Microsatellites can be used to verify that the animals in an inbred colony are essentially pure, with no traces of genetic contamination. This is especially important in facilities that maintain strains with the same coat colour in the same room, a particularly dangerous practice especially when not using individually ventilated cage (IVC) systems.
Microsatellite testing can normally be performed in-house. The number of markers to use for testing has not been standardized: each situation and facility differs in how many and which strains are kept. Nonetheless, a panel of 40 polymorphic SSLPs, evenly distributed across the 19 autosomes, will rule out recent genetic contamination, if the markers can distinguish among the strains being analysed. Alternatively, some institutional core facilities offer SSLP genotyping at reasonable prices.
Interpreting SSLP data is straightforward. Because inbred animals are isogenic and homozygous, they will present only one band in the gel, representing a single allele, when genotyped for a particular SSLP. The presence of any heterozygosity, indicated by two bands, or bands that do not coincide with those of the strain control DNA, should be considered as indicating potential strain contamination (Figure 2). It is important to note that the sizes reported in databases are not always accurate, and that differences in allele size from closely related substrains are possible. Mutations in SSLP loci are rare but possible (and most likely will not produce a phenotype).
Using additional markers flanking the suspect SSLP should resolve the issue and help differentiate contamination from a de novo mutation affecting the SSLP. In the latter case, replacing the breeding colony is not required.
How frequently colony strain identity should be evaluated depends on the size of the  (Figure 3). In addition, hundreds of individual SNP KASP assays for mice and rats are commercially available (e.g., LGC in the UK). This allows customizing the numbers and identities of SNPs for each situation, and avoid using a fixed set of markers that may include non-informative SNPs for a particular application, which cannot be avoided when using SNPs arrays. Another option, real-time PCR (TaqMan®) technology, uses specific primers coupled with a sequence-specific, fluorescent TaqMan probe, is effective and easy to automate; however, the cost per individual assay is expensive compared with KASP assays, and requires a more costly real-time thermocycler.

Discrimination of substrains
The consensus is that if an inbred colony has been genetically isolated for more than 20 generations, it should be considered a substrain, regardless of whether the strain has been confirmed to be genetically different from the parental strain. SSLPs are not recommended for identification of substrains, because there are insufficient numbers of informative markers to compare most of the common substrains. To characterize a novel substrain, it is necessary to use a large set of SNPs. For example, a pairwise comparison of sister strains using the MegaMUGA array showed that the number of polymorphic SNPs is 154 between C57BL/6J and C57BL/6N, 134 between BALB/cJ and BALB/cByJ, and 827 between C3H/HeJ and C3H/HeN 127 . However, only complete exome sequencing, which is becoming more affordable, will provide complete information on mutations that might have occurred in protein coding genes that could influence a specific phenotype. Nevertheless, if the goal is merely to identify the classical substrain a colony or an animal is associated with, then a small number of SNPs can be selected, based on the information available in the SNP databases. As an example, different SNP panels have been proposed to differentiate C57BL/6J from C57BL/6N 89 128 90 . Similar levels of SNP variability were described for several substrains of rats, like F344, SHR, and WKY 92 .

GeMo of outbred colonies
GeMo of outbred stocks is much more complex, because these animals are not genetically uniform. Outbred colonies are essentially a group of closely related animals, with shared ancestors and group identity, but that exhibit some level of genetic heterozygosity. Since outbred colonies form a population rather than a strain, it is difficult to establish a standard GeMo programme with only a few genetic markers. However, with an adequate number of SNPs or SSLPs, allele frequencies within the population could indicate the identity of the stock 17 . One of the main problems of in-house outbred stocks is that they are often maintained with a very small number of animals in the breeding colony, causing a reduction in the number of alleles represented in the population. This may impact genetic drift and increase the inbreeding coefficient. Such colonies are neither truly outbred nor inbred. Although SSLPs or SNPs can be used to estimate the level of heterozygosity within the colony, if it is not possible to keep an appropriate number of breeders, it is better to purchase outbred rodents from vendors that maintain a very large colony and use recommended breeding schemes to reduce inbreeding.

Background characterization (BC) for GA and mutant lines
The explosion in the number of GA lines is exacerbating the problem of undefined "mixed backgrounds" in experimental rodents. This is particularly worrisome for inducible and conditional models that require the crossing of two independent lines Bigenic lines carrying two independent transgenes, two targeted genes, or a combination of both, are also a concern. For example, the Tet-on (or Tet-off) inducible system requires the generation of double transgenic lines carrying the responder and the transactivator constructs in the same animal (Section 2.1).
Likewise, conditional systems such as Cre-LoxP require double transgenic lines that carry both the Cre transgene (or KI locus) and the floxed targeted gene. In both cases, it is very common to see mouse cohorts with mixed backgrounds because the original single-gene lines (e.g., the Cre-expressing line and the floxed line) were from different backgrounds or outbred stocks.
In any case, the problem of mixed background can be circumvented altogether by (i)

Marker-assisted backcrossing for quality assurance and refinement
The use of DNA markers has allowed for a much more rapid and rigorous process of congenic strain development called marker-assisted backcrossing or speed congenics 139 . This process relies on using polymorphic genetic markers covering the whole genome to determine the percentage of donor genome present in the animals, then selecting the animals with the lowest percentage of donor DNA for the next backcross to the recipient strain. This relies on the regions between the polymorphic genetic markers being those of the donor genome: the denser the number of markers the higher the donor genome can be inferred. Common practice is the use of 100-300 markers. This process reduces the number of generations to reach full congenicity (e.g. from N10 to N5), and therefore strain development time, by approximately half.
Using marker-assisted backcrosses and the right number of animals we can obtain ~80% recipient background at N2, ~94% at N3, and ~99% at N4 (compared to the classical mean values of 75.0%, 87.5%, and 93.7%). A flowchart depicting a standard speed congenic protocol is shown in Figure 4. Ideally, the backcross procedure is started with a donor female and a recipient male. Then, F1 mutant males will carry the correct Y-chromosome and after mating to a recipient female, males of the N2 generation will carry the correct X-and Y-chromosome of the recipient strain (avoiding the use of genetic markers on these chromosomes) 140  . This helps address a potential issue with a modifier gene that flanks the genetic change 144 26 . This effort may require breeding many mice to obtain breeders with small introgressed segments.

Genetic stability and cryopreservation programmes
For inbred, co-isogenic, and congenic strains, breeding methods and genetic stability programmes help to minimize substrain divergence due to genetic drift, and also to is best to use a common genetic background (e.g., the floxed strain and the Creexpressing strain on the same background) so that experiments can be replicated.
One recommended scheme for breeding inbred strains is the pyramid-mating scheme. In this system, the foundation colony serves as the genetic and health standard and provides breeders for the top level of the pyramid in every barrier room. This top level, the nucleus colony, is composed of a relatively small number of pedigreed brother-sister mating pairs that produce breeders for the next level of the pyramid, in addition to replenishing itself. In larger colonies, the next level is called the expansion colony, and it provides breeders to the production colony, which in turn produces the animals for experiments. The unidirectional flow of breed stock in this system helps to ensure that any genetic changes or mutations, which would be more likely to occur in the larger expansion or production colonies than in the smaller nucleus colony, will wash out within a single generation 145 .
For outbred stocks, the intent is to minimize inbreeding, maintain heterozygosity and manage genetic drift that would otherwise lead to colony divergence. Ideally, outbred colonies should be maintained with ≥25 breeding pairs, all of which have to contribute to the next generation, in order to avoid an increase of the inbreeding coefficient per generation of more than 1%. Smaller colonies drift fast toward homozygosity because breeders are closely related 146 .
If a new GA line is produced on an undefined or incorrect genetic background, it should be backcrossed for ≥ 4-5 generations to the proper genetic background to describe the phenotype 147 . Subsequently, the line should be cryopreserved, the recovery of viable sperm (through IVF) or embryos after thawing should be tested, and the expected genotypes validated. Embryo freezing has traditionally been the method of choice for archiving mouse lines, while sperm freezing is emerging as more convenient alternative, due to the application of innovative Assisted Reproductive Technologies (ARTs) 148 149 150 . In the near future, archiving and exchanging mutant strains between research facilities will require only obtaining, freezing, storing and transporting mouse sperm 151 152 153 .
Cryopreservation strategies have been adopted for long-term storage of embryos and

Management of rodent facilities
Rodent colony management plans must be carefully considered given the potential for genetic contamination and more classical issues like breeding performance. It is still good practice to separate breeding stocks by coat colour, but comprehensive QA of both the genetic modification and genetic make-up of the strain is imperative.
Standard good practice for colony management should consider breeding matters such as age at mating (e.g., start early at 4-12 weeks of age and replace nonproductive breeders), litter size, inter-litter interval, productive breeding life (e.g., retire breeders after 9 months of age, depending on the strain), pre-weaning mortality rates, and phenotypic issues (e.g. sub-viable life expectancy etc.).

Facility organization
Regarding breeding GA rodents, there will be preferences at individual institutes  Recovery of such models requires the ability to either rederive strains from cryopreserved embryos or through IVF protocols to provide the embryos needed for transfer (it is recommended to seek outside services in case the receiving institution has no embryo technology available). Once recovered, standard animal health checks should be undertaken before releasing the rederived strains into the main breeding or experimental areas of the animal facility/vivarium. Even cryopreservation slows down the genetic drift of the strain, it is important to carefully plan the number of embryos to be frozen, the number of breeding pairs used to produce the embryos (with very clear genetic information) in order to know the genetic background recovered. It is also advisable to keep frozen tissue or DNA to be able to compare the original genetic background has been recovered.

Working according to protocol (Standard Operating Procedures)
A structured approach to the management of colonies is essential to ensure accurate data tracking.  Table 2 for links). When depositing a GA model, care should be taken to provide a technical and breeding profile to ensure receiving institutions can manage their own colonies in a manner consistent with that of the original provider.  This scheme represents the successive steps in the establishment of a congenic strain. The initial step is a cross between the donor strain (albino in the example) carrying the gene of interest (e.g., a targeted gene or a transgene) and a recipient or background strain (black in the example). At each generation, a breeder carrying the gene of interest (*) is backcrossed to a partner of the recipient strain (genetically linked genes are transferred with it and the size of the introgressed fragment can be many Mb, and include many genes). The degree of gray color indicates that, after each backcross generation, the offspring have an increased amount of the background genome (average percentage is indicated in each N generation). When the targeted gene has no easily recognizable phenotype, molecular genotyping is necessary to select the carrier (heterozygous) mice.

Figure 1
Example of genetic contamination detected by SSLP PCR. The picture shows a 4% agarose gel with the characteristic bands obtained after PCR ampliAication using genomic DNA from four mice supposedly belonging to the BALB/c strain (Airst four lanes), plus a standard DNA control for BALB/c (last lane).In this example, only Aive SSLP loci are shown, located in chromosomes 1 to 5. Note the presence of heterozygosity (two bands) and homozygosity for bands that do not match the standard for BALB/c. This is a clear case of loss of authenticity due to genetic contamination. The PCR products are compared with a 100 bp DNA ladder.  Figure 3 Speed Congenic Timeline (~18 months) § Start crossing donor (carrier) female with recipient strain male in order to generate F1 carrier males (PI). § Backcross F1 males to generate ~20 N2 carrier males (~25%) (PI) § Scan N2 carriers with SNPs and select the best breeders (Service) § Cross best N2 males with several recipient strain females (PI) § Generate ~20 N3 carrier males (~25%) (PI) § Genotype N3 mice for heterozygous SNPs in N2 analysis (Service) § Cross best N3 males with several recipient strain females (PI) § Repeat same scheme at N4 (and if necessary at N5)