| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Biochemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706-1544
Correspondence: Address all correspondence and requests for reprints to: Alan D. Attie, Department of Biochemistry, University of Wisconsin-Madison, 433 Babcock Drive, Madison, Wisconsin 53706-1544. E-mail: attie{at}biochem.wisc.edu
 |
Abstract |
|---|
 Top Abstract I. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
|
I. Physiological Definition |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
Hyperglycemia may result from either a relative or absolute insulin deficiency. Absolute deficiency results from a loss of insulin production, e.g., the immunological destruction of ß-cells that occurs in type 1 diabetes. Relative deficiency, as seen in type 2 diabetes, may occur when there is an increased demand for insulin to compensate for a blunted tissue response to the hormone, termed insulin resistance. In this context, insulin production may be insufficient to maintain normoglycemia, despite a relatively robust insulin secretory response. Thus, diabetes can occur in an individual with higher than normal insulin levels. This failure of islets and/or individual ß-cells to maintain sufficient insulin production is an important component of type 2 diabetes in addition to insulin resistance. Thus, in our descriptions of the mouse models below, we have tried to distinguish models of insulin resistance from those that also exhibit ß-cell failure.
|
II. The High Prevalence of Diabetes: Implications for Genetics |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
Approximately one half of the risk for type 2 diabetes is attributable to genetic variation. Twin studies have shown high concordance rates (>80%) for monozygous twins; however the rate is not 100%, demonstrating the effects of environmental factors in the etiology of the disease (6, 7, 8, 9). Individual phenotypes associated with type 2 diabetes, such as ß-cell function and insulin resistance, often have even greater heritabilities (7, 10, 11, 12, 13, 14, 15, 16, 17, 18). Despite the strong genetic component, however, genes with major effects do not account for the bulk of diabetes susceptibility alleles. Together, the genes underlying the rare monogenic forms of diabetes [e.g., maturity onset diabetes of the young (MODY)] account for only a small proportion of the genetic variation associated with type 2 diabetes (7, 19, 20). Instead, numerous genes each contribute a small increase in risk, although if present at high enough frequencies, they may contribute substantially to the population risk, e.g., the Pro12Ala substitution in PPARG (6, 21, 22).
The rapidity of the increase in diabetes onset, coinciding with the adoption of Western lifestyles, suggests that environmental stressors are unmasking prior genetic predisposition. The high incidence of a disorder with a large genetic component suggests that diabetogenic alleles might have experienced positive selection at previous times in human history. The "thrifty gene hypothesis" states that alleles that favor efficient fuel absorption and retention would experience positive selection in times of famine (23, 24). These same alleles would promote diabetes in times of nutritional excess. This hypothesis predicts that regions of the world with populations descended from those having experienced famine will experience a spike in diabetes as they adopt Western-style diets high in caloric density.
Identification of the genetic variants underlying the world diabetes epidemic is tantamount to a genetic-sensitized screen. Millions of people are being exposed to an unprecedented level of metabolic stress in the form of caloric excess. Alleles that were hitherto phenotypically silent are now leading to diabetogenic phenotypes. Although tragic from a public health standpoint, this worldwide gene environment experiment will likely yield new insights into the key bottlenecks in pathways that maintain glucose homeostasis.
|
III. What Does Genetics Offer? |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
The correlation between biochemical measurements and disease suffers from the difficulty in establishing causality. A correlation can occur by chance, it can occur because of a common etiology, or there can be a true cause-effect relationship between the biochemical measure and disease. Genetics offers the power to establish unambiguous models of causality; genetic variation leads to phenotypic variation in a one-way direction. Linkage studies, demonstrating the coinheritance of a specific region of the genome along with the trait of interest, thus require analysis of related individuals and are limited in resolution primarily by the number of individuals. Cross-sectional population-based studies, where large cohorts can more easily be recruited, are necessarily correlational (i.e., showing a correlation between inheritance of a particular marker and phenotype), and do not establish direct causality.
Recent advances in high-throughput single nucleotide polymorphism (SNP) genotyping have greatly increased the markers available for linkage analysis and fine-mapping and are allowing a better understanding of the haplotype structure of the genome (25). With this knowledge, it is possible to identify conserved haplotype blocks inherited from a more distant common ancestor that cosegregate with the trait of interest. As such, whole genome association studies with very dense SNP genotyping (hundreds of thousands of SNPs per individual) are being performed. This will allow interrogation of the genome at a heretofore unprecedented level of resolution. The assembly of very large, well-defined cohorts and collaborative efforts among multiple groups will also aid these efforts and, indeed, is already yielding success (26). These advances suggest that the pace of gene discovery in complex human disease will greatly accelerate. Several recent reviews of human diabetes genetics describe the progress to date (6, 7, 8, 9, 19, 21, 27, 28).
|
IV. Advantages of Model Organisms such as the Mouse |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
Once a linkage region has been identified, one is able to deconstruct a complex trait, breaking it into individual components, through the generation of congenic strains wherein a QTL region from one strain background is introgressed into the other strain background. This creates individuals that are genetically identical except at the locus of interest. For example, a type 2 diabetes-susceptible strain may have separate loci affecting insulin secretion, insulin resistance, and ß-cell survival. Congenics for each of these loci allow each trait to be studied independently. In addition to fine-mapping, this process can also separate multiple QTL lying under the same initial logarithm of the odds (LOD) peak. Nowhere has this been better demonstrated than with the type 1 diabetes susceptibility loci in the nonobese diabetic (NOD) mouse, or type 2 diabetes susceptibility loci in the Goto Kakizaki (GK) rat (29, 30, 31, 32).
For complex diseases like type 2 diabetes, where numerous factors, both genetic and environmental, can affect the trait being studied, the strength of the correlation between variation at a particular gene locus and a phenotype is affected by how far removed the underlying gene is from the phenotype. For example, fasting insulin levels are influenced by insulin sensitivity, insulin clearance, and insulin secretion. Insulin secretion, in turn, is affected by the total ß-cell mass and by the ability of individual ß-cells to sense and respond to glucose. Similarly, numerous processes underlie insulin resistance, and so forth. Thus, although these traits are still complex, they have less genetic heterogeneity than a broader trait, such as plasma insulin levels (14). Studies that focus on these individual components of the phenotype are more likely to detect genetic linkages than studies that merely measure fasting insulin. Mapping sensitivity and fine-mapping may also be increased by studying more detailed phenotypes that are closer to the underlying etiology of the phenotype, thereby reducing the noise from other genetic and environmental impacts on more downstream phenotypes. Extremely detailed phenotypic refinement may even lead to identification of the specific process affected (e.g., glucose sensing vs. vesicle trafficking in models of insulin secretory defects). Animal models offer an advantage in that tissue samples are available for analysis to allow this greater phenotyping capability. In addition, phenotyping is performed in a controlled environment, reducing environmental noise. This also allows for greater flexibility in examining gene-environment (such as gene-diet) interactions on QTL.
The mouse strains together contain as much genetic diversity as the human population; however, much less of this diversity is represented in the commonly used inbred strains (33). Recent SNP analysis suggests that all the commonly used strains are derived from a limited number of founder strains, setting an upper limit to the number of ancestral alleles at any given locus (34, 35). Nonetheless, the haplotype patterns across the strains are complex (36, 37, 38, 39). The strains differ in their susceptibility to disease and in their response to environmental factors like diet. Strain background has been shown to influence most aspects of diabetes including insulin secretion, insulin resistance, and pancreatic characteristics such as ß-cell survival and islet number.
Through the years, many spontaneous mutations have arisen and been preserved as valuable new animal models. The obese (ob) mouse with a mutation in the gene for leptin (Lep) and diabetes with a mutation in the leptin receptor (Lepr) are prominent examples (40, 41, 42). Similarly, other mutations result in spontaneous obesity and severe insulin resistance, such as the fat mutation in carboxypeptidase E (Cpe), the tubby mutation, and the ectopic expression induced by the yellow (y) allele of agouti (Ay) (43, 44, 45, 46). However, the diabetic phenotype associated with these mutations is highly dependent upon the strain background in which they are studied (as described in the sections below). The Akita mouse represents the only spontaneous mutation in mouse models directly causing nonobese type 2 diabetes (47). We have summarized the features of this model later in this review.
Only approximately 300 genes exist in humans that were not identified in the initial mouse genome assemblies (48, 49). There are large regions of synteny between the mouse and human genomes (49), thus linkages are easily translated between genomes. Under contract from the National Institutes of Health, Perlegen, Inc., is sequencing the genomes of 15 commonly used and genetically diverse mouse strains to capture the spectrum of genetic variation that exists within these strains. Additional projects are under way to genotype hundreds of thousands of SNPs in numerous strains to allow for haplotype construction and inference of genotypes of strains other than those being sequenced (35). The Mouse Phenome Database contains quantitative phenotype information for most of the commonly used mouse strains, with the ability to correlate these phenotypes with the available SNP data (50, 51). These valuable resources will greatly aid positional cloning projects in the mouse. A listing of these useful web sites is provided in an appendix to this review.
|
V. Origin of Inbred Mouse Strains |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
Most of the strains in use today derived from four subspecies of the house mouse (Mus musculus) that diverged approximately 1 million years ago: the domesticus subspecies that evolved in Western Europe; the musculus subspecies that evolved in Eastern Europe, Russia, and Northern China; the castaneus subspecies that evolved in Southeast Asia and Southern China; and the molossinus subspecies, which is a hybrid of the castaneus and musculus subspecies, from Japan (34, 53). The genealogies of the commonly used mouse strains have been described, and more recently, detailed strain haplotypes have been inferred from SNP analysis (33, 34, 36, 37, 38, 54, 55, 56, 57). This has confirmed that the majority of inbred strains arose from a limited number of founders, primarily of the domesticus and musculus subspecies (34), and has also demonstrated the complex haplotype patterns that exist across strains (39, 57, 58).
Each strain bears a unique combination of disease susceptibility and resistance loci, with the phenotype observed in each strain due to the composite effects of all these loci. Strains that are not diabetic may harbor diabetes-promoting alleles that are masked in an otherwise protective background, but the effects of these alleles may be observed when placed in the context of other diabetogenic genes. Thus, genetic analysis in animals whose genomes have been combined can reveal diabetogenic alleles contributed by the nondiabetic parental strain. Phenotypic differences across these strains exist for the majority, if not all, phenotypes that have been assessed, and this wealth of diversity has enabled genetic mapping of qualitative and quantitative trait loci controlling these differences.
In Section VI, we describe the genetic and genomic methods that can lead to gene discovery. In Section VII, we summarize the phenotypes of the mouse strains used in diabetes research and the linkages that have been described in these models. We identify examples where specific genes related to diabetes have been identified. We have avoided mention of the numerous insights that have been obtained through transgenic and knockout models, because these have been reviewed extensively elsewhere (59, 60, 61, 62), and we only refer to them where strain has been shown to modify the phenotype.
|
VI. Experimental Strategies for Gene Mapping |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
When a trait is scored in a binary fashion (affected vs. unaffected), then every new individual is informative and helps to narrow the position of the gene responsible for the trait. In an experimental cross for a dichotomous trait, the mapping resolution is simply a function of the sample size. But, for a quantitative trait like blood glucose or insulin, one cannot score animals as affected or unaffected because their values fall on a continuum. The mapping resolution is largely limited by the effect size of the locus (63). QTL often have small effect sizes. Thus, even with infinitely dense markers, mapping resolution is still limited for most quantitative traits. This helps us to understand why only about 1% of genes underlying the more than 2,000 rodent QTL have been identified (64).
B. Backcross and intercross
In mouse genetics, the most common method for gene mapping is to breed together the two inbred strains in which a phenotypic difference has been detected. These F1 offspring are then bred to one or both of the parental stains (backcross) or to each other (intercross) to generate an F2 sample (Fig. 1
). Each time an F1 animal makes gametes, it generates recombinant chromosomes, with each gamete containing a unique distribution of the parental alleles. Thus, a backcross generates mice with discernible products of recombination on the haploid set of chromosomes inherited from the F1 parent, whereas in an intercross between two F1 animals, one can find recombinants on each transmitted chromosomal strand. The offspring of a backcross are either homozygous for one parental genotype or heterozygous. Thus, only alleles acting in a dominant fashion can be detected. In contrast, an F2 has all three possible genotypes at any locus, i.e., homozygotes for each parental allele and the heterozygote, and thus provides information on additivity and dominance. To obtain a picture of all the QTL, an F2 has a higher degree of power, requiring fewer individuals. However, for major dominant QTL, a backcross may have more power (65).
|
C. Recombinant inbred strains
One disadvantage of backcross and F2 samples is that each test animal is unique; thus there is no possibility of biological replication. Recombinant inbred (RI) strains are produced by brother-sister mating of pairs of F2 mice and their offspring (Fig. 2). After 20 generations, each new line of animals is essentially inbred and has fixed in their genomes a unique mosaic of alleles from the two parental strains. Unlike an F2 sample, a RI panel is homozygous at all loci. RI strains were first used in genetic mapping in 1971 (52). Owing largely to the efforts of Benjamin Taylor, The Jackson Laboratory created several panels of RI strains from widely used strains differing in susceptibility to various diseases and infections. Mapping a trait simply requires scoring that trait in the panel of strains and comparing the strain distribution pattern of the trait with that of the marker genotypes. The panel need only be genotyped once, but because each strain is propagated, animals can be phenotyped on multiple occasions across time and even institutions. This allows all data obtained on an RI panel to be correlated to identify common loci affecting multiple phenotypes.
|
A second design that maintains the reproducibility of RI strains but creates more genetic diversity is the RI intercross, which creates F1 animals from each pair of RI strains in the panel (74). Finally, additional backcrossing before inbreeding generates recombinant congenic strains (RCS) in which each strain carries randomly distributed congenic segments of donor genome, inversely proportional in amount to the number of backcrosses performed. For example, each strain from an N3 backcross (an F1 backcrossed twice), will have a random distribution of roughly 12.5% of the donor genome in the recipient background. Several RCS panels have been created, including a reciprocal set between the commonly used A/J and C57BL/6 strains (75).
D. Chromosome substitution (consomic) strains
Chromosome substitution strains (CSS) are created by introgressing an entire chromosome from a donor strain into a background strain. Nadeau and colleagues (76) created a panel of 22 strains that are genetically C57BL/6, except that each contains a single chromosome derived from the A/J strain, which differs from the C57BL/6 strain in many traits. A parallel effort is under way to produce two panels, in which the chromosomes of the Brown Norway strain are each being introgressed into either the Dahl salt-sensitive or the Fawn Hooded Hypertensive rat (77). Lusis and colleagues (78) have created a panel in which various large segments (not entire chromosomes, but overlapping large segments that span chromosomes) of DBA/2 have been introgressed into the C57BL/6 background and have shown that these strains differ in plasma glucose and insulin levels. Genome-tagged mice, in which overlapping segments spanning the entire genome of one strain are introgressed into a different strain have also been created in which DBA/2 or CAST/Ei segments have been introgressed into the C57BL/6 background (79). The power to detect QTL is higher in CSS panels than in an F2 intercross because there is less genetic variance introduced by background loci and because fewer markers need to be tested (80). However, epistatic interactions between donor-derived loci on different chromosomes cannot be detected, and extensive fine-mapping of the resultant loci is required to identify the underlying genetic variant(s).
E. Sensitized screens
A sensitized screen is one in which the animals are rendered more sensitive to a stress by some prior manipulation, and then the response of the animals to the stressor is mapped. In diabetes, this is commonly performed by feeding normally non- or mildly insulin resistant animals high-fat or high-carbohydrate diets, or by studying the response of the animals to a genetic defect that induces insulin resistance, to identify alleles that modify this phenotype. The most commonly used models have been the monogenic obesity mutations, obese (ob) and diabetes (db) in the leptin (Lep) and leptin receptor (Lepr) genes, respectively. The classical studies of Coleman and Hummel (81) were the first to show that strain background affects diabetes susceptibility in mice.
F. Outbred stocks
A powerful alternative to crosses between inbred strains is the generation of outbred stocks (Fig. 3). In analogy to human populations, many generations of outbreeding yield a mosaic that is substantially more genetically fragmented than an F2 intercross. One set, consisting of eight founder strains, was bred for more than 60 generations and now has an average recombination distance of just 1.7 cM (82). An association study can be performed on a sample of 1–2,000 mice derived from this stock and have a power to detect a QTL with a resolution of 0.5–1.0 cM. These studies require dense genotyping, but this is now being facilitated by various automation schemes and genomewide SNP panels. This is similar to an association study in a human population, with the added advantage of knowledge of the haplotype structure of the ancestral strains. An elegant variation of the association study is to cross outbred animals to an inbred strain to create an outbred-inbred F2 (83). Mapping can then be done in two steps, first at low-resolution with a small number of markers, then at higher resolution in selected regions of the genome showing evidence of linkage to the trait. Recent success with this approach has come in the positional cloning of a QTL for anxiety (84).
|
|
VII. Mouse Strains Used in Diabetes Research |
|---|
|
TopAbstractI. Physiological DefinitionII. The High Prevalence...III. What Does Genetics...IV. Advantages of Model...V. Origin of Inbred...VI. Experimental Strategies for...VII. Mouse Strains Used...VIII. Linkage Hot SpotsIX. Summary/ConclusionX. Appendix: Useful Web...References |
|---|
A. C57BL/6
The C57BL/6 strain was created by Little in 1921 from a mating of female no. 57 and male no. 52 from Lathrop’s stock. SNP analysis indicates that the C57BL/6 strain shares some Mus musculus molossinus genomic regions and is somewhat removed evolutionarily from the other inbred strains, except the other C57-derived strains and C58 (36, 37, 88). The C57BL/6 strain is the most widely used of all inbred strains, accounting for some 14% of all such uses. Several RI panels have been generated with C57BL/6 as one strain (72). C57BL/6 is the most commonly used reference strain in studies of diabetes.
There is some confusion about whether or not one should consider the C57BL/6 strain to be "diabetes-susceptible" or "diabetes-resistant." Both protective and diabetes susceptibility alleles/phenotypes have been observed in this strain. Thus, the phenotype is largely relative to the strain it is being compared with and may depend on the criteria by which diabetes has been defined. C57BL/6 may be best described as a strain of intermediate susceptibility. Indeed, data from the Mouse Phenome Database shows that glucose and insulin levels of this strain are intermediate to those of the other strains (2).
Several studies have provided evidence of increased diabetes susceptibility in the C57BL/6 strain. In parallel to studies of diet-induced atherosclerosis susceptibility, the C57BL/6 mouse emerged as susceptible to diet-induced obesity and hyperglycemia (89, 90, 91, 92). Studies from the Surwit laboratory showed that a diet enriched with fat and carbohydrate (mainly from disaccharides) produced a roughly 50% increase in fasting plasma glucose and a 10.4-fold increase in plasma insulin in C57BL/6 mice (91). By contrast, glucose remained nearly unaltered, and insulin rose only 2.6-fold in A/J mice, demonstrating enhanced diet-induced insulin resistance and hyperglycemia in C57BL/6. Kaku et al. (93) found C57BL/6 to be the least glucose tolerant of six strains fed a normal chow diet. More recently, Goren et al. (94) compared glucose and insulin tolerance of C57BL/6, C57BLKS/J, DBA/2, and 129X1. Consistent with the aforementioned study of Kaku et al. (93), C57BL/6 was inherently less glucose tolerant in a glucose tolerance test. But, in an insulin tolerance test, the two C57 strains were more sensitive to the exogenously added insulin than DBA/2 or 129X1, suggesting that a defect in insulin secretion is the likely cause of the glucose intolerance of the C57BL/6 mice. An in vivo study that assessed insulin sensitivity and insulin secretion during the early and late stages of feeding a high-fat diet to C57BL/6 mice concluded that after longer-term feeding, although total insulin secretion was sufficient to compensate for insulin resistance, defective acute insulin secretion in response to glucose is a major contributor to glucose intolerance (95).
Other investigators have also identified an insulin secretion defect in the C57BL/6 strain compared with other strains (93, 96, 97, 98, 99). In perifusion experiments, islets from high-fat, high-simple carbohydrate diet-fed C57BL/6 mice had markedly impaired secretion of insulin in response to high (27.7 mM) glucose compared with A/J (100, 101). This was primarily attributed to a defect in the second-phase insulin secretion. Furthermore, C57BL/6 islets have been shown to be unresponsive to intermediate glucose levels (10 mM), due to incomplete closure of their KATP channels (98).
These studies are consistent with the C57BL/6 mouse having a ß-cell defect in glucose responsiveness, which can, to a certain point in vivo, be offset through the generation of sufficient ß-cell mass. In vivo, ß-cells from this mouse strain have very high proliferative capacity (102). A study by Bock and co-workers surveyed the pancreatic architecture of several strains. In comparison to the others, lean C57BL/6 have a relatively low total islet mass and ß-cell mass when expressed per body mass. But, C57BL/6 had the highest number of islets, albeit the smallest islets, consistent with a large foundation from which ß-cell mass can expand when stressed (103). However, this relatively small existing ß-cell pool may also contribute to the observations that C57BL/6 is more sensitive than other strains of mice to the ß-cell destructive effects of streptozotocin (104, 105).
To identify genes involved in the glucose intolerance of the C57BL/6 strain, Toye et al. (98) created an F2 intercross from C57BL/6 and C3H/HeH. The mice were phenotyped for their response to glucose and insulin tolerance tests. Fasting insulin, postglucose insulin, and postinsulin glucose linked to distinct loci, confirming that variation in each of these parameters has a genetically distinct cause.
A QTL on chromosome 13 that was linked to postglucose insulin levels contains the nicotinamide nucleotide transhydrogenase (Nnt) gene. Sequencing of the cDNA for this gene revealed a M35T polymorphism within the mitochondrial leader sequence. The C57BL/6 allele also has an in-frame deletion of exons 6 through 11, leading to a transcript of approximately 508 instead of approximately 1,261 bp. The gene is expressed at a much lower level in C57BL/6 than in C3H/HeH islets. The syntenic region to the mouse Nnt locus on human chromosome 5p15 is one of three loci linked with a genetic modifier of age of onset of diabetes in families with HNF1
(MODY3) mutations (106).
Nnt is an inner mitochondrial membrane enzyme that catalyzes the exchange of reducing equivalents between nicotinamide-adenine dinucleotide (NAD+) and NAD phosphate (NADP+). The reaction is coupled to the transport of protons from the cytosol to the mitochondrial matrix: NADH + NADP+ + H+cytosolic 
NAD+ + NADPH + H+matrix.
Together with the QTL mapping study, a very strong case implicates Nnt as the gene responsible for impaired insulin secretion in the C57BL/6 mouse. Knockdown of Nnt expression in the Min6 insulinoma cell line abolished the insulin secretion normally observed in response to 10 mM glucose (107). The same group screened an N-ethyl-N-nitrosourea (ENU)-mutagenized DNA archive and found gametes from two mutant mice with different Nnt nonconservative missense mutations, one in the NAD+ binding domain and the other in a transmembrane domain. Mice expressing either of these mutant alleles had impaired glucose tolerance, and islets from these mice had impaired glucose-stimulated insulin secretion. In short, the mutant mice reproduced the insulin secretion phenotype of the C57BL/6 mouse. Finally, transgenic expression of the 129S6/SvEvTac allele of the Nnt gene rescued the insulin secretion phenotype of the C57BL/6 mouse (108).
It is noteworthy that this gene was identified and implicated in insulin secretion through genetic linkage rather than a biochemical hypothesis. How would loss of Nnt function reduce insulin secretion in response to glucose? Islets harboring the mutant Nnt allele produce a much higher amount of reactive oxygen species in response to 20 mM glucose (107), supporting the prior hypothesis that Nnt detoxifies reactive oxygen species (109). Because reactive oxygen species have been shown to increase uncoupling of oxidative phosphorylation, Nnt would be predicted to indirectly affect mitochondrial ATP yield and thus the extent of the closure of the ATP-sensitive K+ channels. This prediction is consistent with the reduced glucose utilization seen in islets from the mutant mice (107).
Another way to metabolically stress mice to find strain differences in diabetes susceptibility is to induce insulin resistance by introgressing null alleles in genes required for insulin signaling. When C57BL/6, DBA/2, and 129X1/Sv mice were created that were double heterozygotes for null alleles at the insulin receptor and insulin receptor substrate loci, all developed severe insulin resistance, as expected (110). The C57BL/6 mice developed glucose intolerance and dramatic hyperinsulinemia, which correlated with a striking 22-fold increase in ß-cell mass. The DBA/2 mice also developed hyperinsulinemia, but did not have as severe glucose intolerance or as strong a ß-cell proliferative response as the C57BL/6 mice. Thus, it was the C57BL/6 mice that had a much higher prevalence of diabetes.
Despite the decreased insulin secretion and glucose intolerance of the C57BL/6 strain, C57BL/6 mice are relatively resistant to obesity-induced diabetes when compared with several other strains, as described below. This was revealed in the classic experiments of Douglas Coleman conducted over 30 yr ago. His experiments were the first to show that strain background affected the susceptibility of mice to obesity-induced diabetes. Coleman found that genetically obese mice (due to homozygosity for the ob mutation in the leptin gene; Lepob/ob) in the C57BLKS/J strain background are severely diabetic, whereas genetic obesity in the C57BL/6 background was associated with severe insulin resistance, but not the development of diabetes (111). In the C57BL/6 background, homozygosity for the Lepob mutation led to moderate (
220 mg/dl), transient hyperglycemia between 8 and 12 wk of age, but then glucose returned to near normal levels (81). This return to normal glucose levels is due to a marked increase in insulin arising from a dramatic expansion of ß-cell number in the C57BL/6 mice that is not sustained in the C57BLKS/J background (112), consistent with the aforementioned high ß-cell proliferative capacity of C57BL/6 mice. C57BL/6 mice are also relatively resistant to the development of diabetic nephropathy (113).
B. C57BLKS/J
The C57BLKS/J strain arose through the inadvertent genetic contamination of C57BL/6 (114). Although the non-C57BL/6 alleles are scattered across all the chromosomes and are primarily from the DBA/2 strain, recent SNP analysis indicates that there are contributions from at least three other strains, including C57BL/10, 129, and another unidentified strain that accounts for nearly 10% of the genome (36, 115, 116). As described above, the diabetes severity associated with homozygosity for the Lepob or Leprdb mutations is much greater on the C57BLKS/J than the C57BL/6 background (81). Because the genomic constitution of this strain derives primarily from C57BL/6 and this strain is more diabetes-prone than C57BL/6, we can presume that the diabetogenic alleles are derived from the contaminating strains or perhaps the interaction between these alleles and diabetogenic C57BL/6 alleles.
Lean C57BLKS/J are very sensitive to streptozotocin (117). C57BLKS/J mice have an approximate 50% reduction in ß-cell proliferative capacity compared with C57BL/6, and an increased ß-cell susceptibility to glucotoxicity (118, 119). Mean islet area is increased in male C57BLKS/J mice, but total pancreatic insulin content is reduced compared with C57BL/6 (119).
Mapping studies have been performed in C57BL/6 x C57BLKS/J Leprdb/db F2 mice. Coleman found concordance between the segregation of a malic enzyme regulatory locus on chromosome 12 and diabetes in a Leprdb/db F2 between C57BL/6 and C57BLKS/J, with the low activity allele associated with diabetes (111, 120). More recently, Paigen and colleagues (121) identified additional suggestive loci for increased glucose levels from C57BLKS/J alleles. A mapping study on C57BLKS/J x DBA/2 Leprdb/db F2 mice identified a locus on chromosome 9 associated with impaired glucose tolerance (122). The DBA/2 allele at this locus contributes to the increased glucose levels, suggesting that the C57BLKS/J allele is a C57BL/6-derived protective modifier and that the DBA/2 strain has diabetogenic loci in addition to those introgressed into the C57BLKS/J strain.
It is unclear, however, whether the effects of these alleles require the absence of effective leptin signaling. When the fat mutation in carboxypeptidase E (Cpe), which was originally identified on the HRS strain background, was introgressed onto the C57BLKS/J background, C57BLKS/J-Cpefat/fat males became diabetic despite reduced adiposity, whereas HRS-Cpefat/fat mice were more hyperinsulinemic and did not develop hyperglycemia (123). However, C57BLKS/J-Cpefat/fat mice were more insulin sensitive and showed less ß-cell/islet damage than C57BLKS/J-Leprdb/db mice (124). In fact, the hyperglycemia reached a plateau and eventually reversed in these animals, suggesting that the severe diabetes in C57BLKS/J-Leprdb/db mice results from an interaction between C57BLKS/J alleles and the Leprdb mutation. This could in part explain why high-fat, high-sucrose feeding showed C57BLKS/J to have a reduced response in terms of weight gain and glucose and insulin levels compared with C57BL/6 (125). A recent study in C57BLKS/J x DBA/2 F2 mice, either Leprdb/db or Leprdb/+, has shown that regulation of glucose levels is subject to complex interactions between sex and Leprdb genotype and epistatic interactions between loci (126).
C. DBA/2
The DBA strain was the first inbred strain created by Little, in 1909, deriving its name from its coat-color alleles dilute (d), brown (b), and non-agouti (a). Of the two substrains, it is the DBA/2 substrain that is commonly used in diabetes research. Given the strain differences between C57BLKS/J and C57BL/6 in diabetes susceptibility and the knowledge that C57BLKS/J and C57BL/6 differ genetically primarily due to the introgression of DBA/2-like alleles, this strain has been used as a model for genetic studies as well.
Like C57BLKS/J-Leprdb/db, DBA-Leprdb/db mice become severely diabetic (111). DBA/2 Lepob/ob mice are also severely diabetic; however, unlike C57BLKS/J obese mice, not all the mice undergo pancreatic decompensation (127). Because the C57BLKS/J strain is primarily a combination of the C57BL/6 and DBA/2 genomes, this suggests that either non-DBA/2 non-C57BL/6 alleles in C57BLKS/J or interactions between DBA/2 and C57BL/6 alleles promote the severe decompensation seen in the C57BLKS/J strain. However, as described above, there are also likely additional diabetogenic alleles in the DBA/2 strain not present in C57BLKS/J (122, 126). The DBA/2 strain is also susceptible to diabetic nephropathy (113).
In contrast to genetic obesity, lean DBA/2 mice that are fed a high-fat diet are more glucose tolerant than C57BL/6 (98). In a survey of nine inbred strains, DBA/2 had the lowest glucose levels on high-fat feeding (92), and data from the Mouse Phenome Database show that DBA/2 mice have among the highest insulin levels (2). When insulin resistance is induced through double heterozygosity for insulin receptor (Insr) and insulin receptor substrate-1 (Irs1) knockout alleles, DBA/2 mice have reduced insulin resistance and a slower onset of hyperglycemia than C57BL/6 (110). DBA/2 mice have higher insulin levels in the fed state than C57BL/6 and other strains (93), which suggests that DBA/2 may have increased insulin secretion. Indeed, compared with C57BL/6, DBA/2 hypersecrete insulin in response to glucose from birth, which has been attributed to increased islet glucose utilization and GLUT2 levels (97). In a survey of pancreatic structure across seven mouse strains, DBA/2 also had the most islet and ß-cell mass of all the strains (103).
Combined, these data suggest that DBA/2 ß-cells are acutely sensitive to obesity. The DBA/2 strain is genetically distinct from the other strains used in diabetes research (36, 37) and may provide important insight into factors underlying obesity-induced ß-cell failure. A set of very large insert congenic strains spanning the genome in which DBA/2 alleles have been introgressed into the C57BL/6 background have been developed (78) and will be an excellent resource for mapping these differences.
D. 129
The 129 strain has a complicated breeding history, which has resulted in numerous distinct substrains in existence today. Outcrossing among these substrains has led to several genetic differences between them, which have been extensively characterized, and a standardized nomenclature has been developed (128, 129). The majority of ES cell lines used in generating gene-targeted mice are derived from substrains of the 129/Sv strain, a heterogeneous strain which is the ancestor of many of today’s substrains (128, 129). Thus, most knockouts are congenic strains in which the targeted allele and some amount of flanking sequence have been introgressed into other strain backgrounds. Depending on the amount of remaining flanking sequence, it is possible that nearby 129/Sv-derived alleles may contribute to the observed phenotype. For simplicity, here we refer to all substrains as 129, however it should be noted that the phenotypic effects may differ between the various substrains.
On both chow and high-fat-diet feeding, the 129 strain maintains low insulin and is more glucose tolerant than other strains (2, 130, 131). This suggests that 129 mice have increased insulin sensitivity and is consistent with the observation that wild-type 129 mice have low ß-cell mass (103, 110). In a C57BL/6x 129 F2 study, five loci for hyperglycemia or hyperinsulinemia have been identified (130). At each of these loci, the 129 allele was associated with the lower insulin or glucose phenotype.
Thus, the 129 strain is diabetes-resistant in the context of genetic obesity. Alleles from the 129 strain are capable of conferring protection to C57BLKS/J- Leprdb/db mice (132). Similarly, mice homozygous for a db allele (db3J) found on this strain have very mild and transient hyperglycemia, marked hyperinsulinemia, and dramatically elevated pancreatic insulin content. Although this is accompanied by progressive hyperglucagonemia, the mice ultimately develop hypoglycemia leading to sudden death (133).
Similarly, lean 129 mice with genetically induced defects in insulin signaling due to double heterozygosity for Irs1 and Insr mutations have only mild hyperinsulinemia with little insulin resistance and do not increase ß-cell mass compared with wild-type 129 mice (110). This is in contrast to C57BL/6 and DBA/2 mice bearing the same mutations, which develop marked hyperinsulinemia with much more severe insulin resistance and resulting in diabetes (110, 131). In these double-heterozygous mice, linkages for plasma glucose levels have been identified on chromosomes 12 and 14, where C57BL/6 alleles are diabetogenic (131).
In contrast to the insulin resistance evoked by Irs1-Insr double heterozygosity, heterozygosity for only a null allele of the Insr results in more severe hyperinsulinemia in the 129 than C57BL/6 background (134). When loci were mapped in an Insr+/– C57BL/6x 129 F2 population, five loci (significant and suggestive) affecting plasma insulin levels were identified; however, at four of these loci, C57BL/6 alleles were linked to the higher serum insulin concentrations. Nonetheless, this suggests that the 129 strain may harbor at least one diabetogenic locus.
Additional evidence of diabetes promoting alleles in the 129 strain come from other knockouts in the insulin signaling pathway. The insulin receptor substrate-2 (Irs2) gene is important for the normal survival and development of ß-cells (135). In a C57BL/6x 129/sv mixed background, Irs2–/– mice have reduced Pdx1 expression and develop diabetes (136). However, mice with this mutation on the pure C57BL/6 background preserve the expression of the Pdx1 transcription factor and its downstream genes (137). This suggests that alleles from the 129 background contribute to the down-regulation of Pdx1.
E. BTBR
The black and tan, brachyuric (BTBR) mouse strain was developed by L. C. Dunn from an outbred stock he obtained from Nadine Dobrovolskaia-Zavadskaia (138). The strain was maintained by Dorothea Bennett after 1962, and then by her graduate students, Flaherty and Artzt, who passed it on to Jean-Louis Guenet at the Pasteur Institute. Guenet maintained the strain for 45 generations, then passed it to Dove’s laboratory at the University of Wisconsin. During its breeding history, it was bred with 129 mice and selected for good breeding performance (139). Thus, it is genetically most related to the 129 strain (36, 57). However, in contrast to 129, BTBR is a diabetes-susceptible strain.
Lean male F1 hybrids between C57BL/6 and BTBR are more glucose intolerant than either parental strain, suggesting that interactions between C57BL/6 and BTBR alleles can further promote diabetes susceptibility (140). Insulin-stimulated glucose uptake into isolated adipocytes from BTBR (141) and the BTBR x C57BL/6 F1 mice is severely defective (142). Isolated skeletal muscles from the F1 mice also have defective insulin-stimulated glucose transport, although a much milder defect is observed in muscles from lean BTBR mice (140). In vivo, glucose uptake into muscle and adipose tissue in male BTBR mice under hyperinsulinemic clamp conditions is less than C57BL/6, although insulin-mediated suppression of hepatic glucose output is unaffected in BTBR (J. B. Flowers and A. D. Attie, unpublished observations). Islets isolated from BTBR mice hypersecrete insulin at low glucose concentrations and in response to nonglucose secretagogues (143).
Similar to C57BLKS/J mice, when the Lepob mutation is bred into BTBR background, BTBR Lepob/ob mice develop severe diabetes (141, 144). Young BTBR Lepob/ob mice are hyperinsulinemic compared with C57BL/6 Lepob/ob. However, as these mice age, insulin levels decline, resulting in marked hyperglycemia (141). By 3 months of age, pancreatic insulin content is dramatically reduced, and islet architecture is severely disrupted in BTBR Lepob/ob mice (our unpublished observations). We have used this model for several genetic and genomic studies (144, 145, 146, 147) and in a cross between BTBR and C57BL/6 mice, we have mapped several diabetes-related QTL in Lepob/ob F2 animals (144).
Using overlapping interval-specific congenic (65) and subcongenic strains, Clee et al. (148) recently positionally cloned one of these loci on chromosome 19. The gene underlying the t2dm2 QTL was found to be SorCS1 (148), a receptor expressed in islets that binds to platelet-derived growth factor-BB (149). Platelet-derived growth factor-BB is required for pericyte recruitment to the islet microvasculature, a step critical for vessel formation and stability (150, 151). Thus, we believe that SorCS1 may play a role in the development or maintenance of the islet vasculature, which would be important for expanding ß-cell mass in response to insulin resistance. Interestingly, the diabetogenic allele at this locus was contributed from the C57BL/6 strain (144, 148) and was associated with reduced insulin secretion in vivo, suggesting that alterations in SorCS1 may also contribute to the impaired insulin delivery to the bloodstream of the C57BL/6 strain.
F. A/J
The A/J strain is the classic "diabetes-resistant" mouse strain and is among those with the lowest glucose levels (2). When fed a high-fat, cholesterol-containing diet, A/J mice maintain very low glucose and insulin levels. Compared with the C57BL/6 strain, A/J mice are resistant to high-fat, high-carbohydrate feeding-induced obesity, insulin resistance, and glucose intolerance (89, 90, 91, 125), in part due to increased insulin secretion and a sustained second-phase response (100, 101). Thus, A/J mice have reduced glucose even on chow (91, 152) and do not expand ß-cell mass when fed high-fat diets (100). However, even this resistant strain may harbor some alleles promoting glucose intolerance (153, 154, 155). It is a common progenitor in RI and recombinant congenic panels.
G. BALB/c
The BALB/c strain is genetically most related to A/J (36, 37). Consistent with this, it is more glucose tolerant than C57BL/6 (98) and more resistant to the glucotoxic effects of streptozotocin (104, 117). BALB/c mice have a relatively high ß-cell mass due to a large number of islets (103).
In the BALB/c background, Lepob/ob mice have increased insulin levels in the fed state compared with C57BL/6 Lepob/ob (156). Interestingly, BALB/c Lepob/ob mice have reduced adiposity, which correlates with increased thermogenesis. In contrast to other strains, BALB/c mice homozygous for the Lepob mutation remain fertile (156).
H. C3H
C3H/HeJ is a diabetes-resistant strain. This strain is related to BALB/c, A/J, and AKR (36). It is highly glucose tolerant and has a robust insulin secretory response, which is sufficient to overcome its slight insulin resistance (93, 98). C3H mice are resistant to the ß-cell toxic effects of streptozotocin, possibly due to the effects of a single gene (105, 117). ß-Cell mass in C3H is derived from relatively few islets of large size (103), resulting in increased ß-cell mass compared with C57BL/6, but less than other strains such as DBA/2 and BALB/c. As described above, mapping in F2 and F3 samples derived from C3H and C57BL/6 has identified loci on chromosomes 2 and 13 associated with improved glucose tolerance relative to C57BL/6 (99). In contrast, a C3H-derived allele associated with impaired glucose tolerance has recently been mapped in a cross with C57BLKS/J-db mice (157).
I. AKR
AKR mice are closely related to the LG strain, described below, and are also related to C3H, BALB/c, and A/J (36). The AKR strain is very sensitive to diet-induced obesity (92), and adipocytes from this strain exhibit increased insulin-stimulated glucose uptake compared with SWR/J mice (158). Unlike its related strains described above, when fed a high-fat diet, AKR mice are more hyperinsulinemic than others (2). They are more insulin resistant than C57BL/6, primarily due to reduced glucose uptake into adipose tissue (96). Despite this, AKR are more glucose tolerant and have lower plasma glucose levels than C57BL/6 because of increased insulin secretion during a glucose challenge (96). AKR mice are relatively resistant to the effects of streptozotocin (117).
J. CAST/Ei
The CAST/Ei strain is an inbred line derived from wild Mus musculus castaneus mice from Thailand. Compared with other strains with data in the Mouse Phenome Database, CAST/Ei mice have relatively high glucose levels (2). The Lusis group (159) has made extensive use of this strain in studies of obesity and atherosclerosis. CAST/Ei mice are very lean, but, on a high-fat diet, hyperinsulinemic relative to C57BL/6. In an F2 derived from CAST/Ei and C57BL/6, three independent loci on chromosome 2 segregate with body weight loci and exhibit suggestive linkage to plasma insulin levels (159). Differences in glucose and insulin levels have been observed in congenic strains where CAST/Ei alleles have been transferred into the C57BL/6 background (160). In a recent study, a new selectively inbred strain, the Horio-Niki diabetic (HND) mouse, was derived from C57BL/6 and M. m. castaneus parents (161). These nonobese mice have impaired insulin secretion in vivo, and an altered response to 2-adrenergic stimulation, suggesting that this may be a model of impaired neurological control of insulin secretion (162).
K. Nonobese diabetic (NOD) and nonobese nondiabetic (NON)
These strains have a complicated history. Originally, a mouse with spontaneous development of cataracts was observed in a closed outbred colony of Imperial Cancer Research (ICR) mice. This mouse was selected and inbred to develop the CTS (Cataract-Shiongi) line, however none of the offspring developed cataracts (29, 163). As cataracts can be a complication of diabetes, at the F6 generation, two lines were split, one with euglycemia (A) and one with mild hyperglycemia (B). At the F20 generation, a spontaneously diabetic mouse was observed in the euglycemic A line. The offspring of this mouse were inbred for several generations with selection for spontaneous diabetes to derive the NOD line, whereas mice from strain B did not develop spontaneous diabetes, and became the NON control line (163).
The NOD strain is a model of type 1 autoimmune diabetes, with a complex genetic etiology. The genetics of this strain have been extensively studied and involve at least 20 different loci, none of which independently result in diabetes (29). The genetic complexities of this strain are also evident in that several of the mapped Idd loci have ultimately been dissected into multiple distinct loci (29, 164). Analysis of these loci often involves haplotype comparisons across several diabetes susceptible and resistant strains, in congenics where the genotype at several other loci is controlled (164, 165). Interestingly, alleles that further promote diabetes in NOD mice derived from the nondiabetic control strains, e.g., C57BL/6, have also been observed (166), and alleles associated with the sensitivity of this strain to streptozotocin have been mapped (167). We will not focus on the models of type 1 diabetes in this review, but point the reader to other reviews of the subject (168, 169, 170, 171, 172).
Of more relevance to type 2 diabetes is the NON strain, which has been shown to have alleles promoting type 2 diabetes (173, 174, 175). This should not be particularly surprising, given its selection for mild hyperglycemia during its breeding history and that several other models of ß-cell dysfunction and type 2 diabetes have been derived from the outbred ICR mice (see descriptions of the NSY and ALR and ALS strains below), suggesting that this stock has numerous alleles promoting diabetes (29, 163). As the genes are identified, it will be interesting to observe whether there is any overlap between type 1 and type 2 diabetes susceptibility alleles (29). NON mice are glucose intolerant and have impaired glucose-stimulated insulin secretion, but they do not transition to overt diabetes even when fed an 11% high-fat diet (176). With 45% fat and simple carbohydrates, the mice become transiently hyperglycemic and very obese, but eventually go into remission from the hyperglycemia (E. H. Leiter, personal communication).
The genetic mapping studies that mapped diabetes in the NON strain used an F2 intercross between this strain and NZO, described below. In contrast to either parental strain, F1 mice exhibit severe fasting hyperglycemia, and nearly 100% of males develop diabetes (173). Two NON-derived diabetogenic loci were identified in the F2 sample (173), and a third suggestive NON-derived locus was identified in a (NZOxNON) x NON backcross (174).
L. New Zealand obese (NZO)
The New Zealand obese (NZO) strain was derived from an outbred stock from the Imperial Cancer Research (ICR) laboratory in England. It was originally selected for agouti coat color, along with the New Zealand Black (NZB) and New Zealand White (NZW) strains (177), and is a model of spontaneous polygenic obesity and insulin resistance. Leptin receptor polymorphisms have been detected in this strain; however, they do not appear to directly cause the obesity of the strain, because the related NZB strain shares these polymorphisms and is not obese (178). Furthermore, studies mapping obesity-related genes in this strain have not observed linkage to this region (179, 180). It is possible, however, that these polymorphisms may confer leptin resistance or impaired leptin transport across the blood-brain barrier (181) and interact with other obesity-promoting alleles to enhance the obesity of these mice (178, 180).
The phenotype of NZO mice resembles the metabolic syndrome (182). Compared with the related NZB and NZC strains and the lean Swiss Jackson Laboratory (SJL) strain, NZO mice are markedly hyperinsulinemic, with males developing hyperglycemia (182, 183). The mice have reduced insulin-stimulated glucose uptake into muscle and adipose tissue, along with increased hepatic glucose production (183). Serum triglycerides and blood pressure are also increased in these mice (182). First-phase insulin secretion in response to a glucose bolus is impaired in NZO mice (183). After initial ß-cell hyperplasia, decompensation in males is accompanied by reduced ß-cell mass, degranulation, and ultrastructural changes (184). However, in various NZO substrains, not all NZO males decompensate, apparently needing to exceed a body weight threshold before diabetes develops (173, 174, 176).
Like NZB mice, NZO mice have an autoimmunity phenotype; they have autoantibodies against the insulin receptor, which may contribute to their insulin resistance (185). Leukocyte infiltration of the pancreas has also been observed in NZO mice, and diabetes is worsened in NZO x NZB F1 mice (177, 184). However, later studies have shown that this autoimmunity is not essential for the development of diabetes in these mice (176).
Loci contributing to obesity and diabetes susceptibility in this strain have been detected on nearly every chromosome (175). Genetic mapping studies have also been performed in a (NZO x SJL) x SJL backcross panel (179). Interestingly, these studies identified a major locus on chromosome 4 from the SJL background that contributes to the development of diabetes (hypoinsulinemia and hyperglycemia) in these obese mice (179, 186). This locus, Nidd/SJL, is associated with hyperglycemia and relative hypoinsulinemia, due to a loss of ß-cells and islet disorganization in high-fat-fed mice (186). A strong interaction between this locus and one controlling obesity accelerates the progression of diabetes in these mice. Subsequent mapping in female mice demonstrated that NZO-derived alleles contribute to obesity and hyperinsulinemia (180).
In an F2 derived from NZO and NON, Leiter et al. (173) identified three NZO-derived diabetogenic alleles. Two of these were suggestive loci affecting blood glucose concentrations. Interestingly, they were dependent on the direction of the original cross, i.e., whether the parental female or male was NZO. A second study, in an F1 x NON backcross, identified a major NZO-derived diabetogenic QTL (174). This locus influenced body weight, plasma glucose, and early hyperinsulinemia. This study also detected complex interactions between various loci (e.g., those affecting obesity and plasma glucose levels), which may contribute to the increased diabetes in F1 males compared with either paren
25). Maximum genetic diversity is maintained if each breeding male and female contributes a single male and female to the next breeding generation.