Animal models of diabetes mellitus


  • D. A. Rees,

    1. Department of Medicine, University of Wales College of Medicine, Cardiff, UK
    Search for more papers by this author
  • J. C. Alcolado

    Corresponding author
    1. Department of Medicine, University of Wales College of Medicine, Cardiff, UK
      Dr J. C. Alcolado, Department of Medicine, University of Wales College of Medicine, Heath Park, Cardiff, CF14 4XN, UK. E-mail:
    Search for more papers by this author

Dr J. C. Alcolado, Department of Medicine, University of Wales College of Medicine, Heath Park, Cardiff, CF14 4XN, UK. E-mail:


Animal models have been used extensively in diabetes research. Early studies used pancreatectomised dogs to confirm the central role of the pancreas in glucose homeostasis, culminating in the discovery and purification of insulin. Today, animal experimentation is contentious and subject to legal and ethical restrictions that vary throughout the world. Most experiments are carried out on rodents, although some studies are still performed on larger animals. Several toxins, including streptozotocin and alloxan, induce hyperglycaemia in rats and mice. Selective inbreeding has produced several strains of animal that are considered reasonable models of Type 1 diabetes, Type 2 diabetes and related phenotypes such as obesity and insulin resistance. Apart from their use in studying the pathogenesis of the disease and its complications, all new treatments for diabetes, including islet cell transplantation and preventative strategies, are initially investigated in animals. In recent years, molecular biological techniques have produced a large number of new animal models for the study of diabetes, including knock-in, generalized knock-out and tissue-specific knockout mice.


bio breeding


embryonic stem


Goto Kakizaki


insulin receptor substrate 1


insulin receptor substrate 2


maturity-onset diabetes of the young


non-obese diabetic




Otsuka Long-Evans Tokushima fatty rat


Animal experimentation has a long history in the field of diabetes research. The aim of this article is to review the commonly used animal models and discuss the recent technological advances that are being employed in the discipline. The review is based on an extensive literature search using the terms rodent, mouse, rat, animal model, transgenics, knockout, diabetes and pathogenesis, in scientific journal databases such as medline®. In addition, abstracts presented at meetings of Diabetes UK, the European Association for the Study of Diabetes and the American Diabetes Association over the last 5 years were examined in order to gain an appreciation of recent and ongoing research projects.

 The topic is vast and this review is necessarily limited in its depth, intending to cover models of both Type 1 and Type 2 diabetes. Nevertheless, it should serve the purpose of updating practising clinicians and those not currently active in this type of research regarding the approaches that are currently being used. It should also serve as a starting point for those wishing to embark on an animal-based research project. Interested readers are pointed to the numerous specific reviews cited throughout this article and the excellent text ‘Animal Models of Diabetes: A Primer’[1].

Ethical, moral and legal issues

At the outset, it is important to acknowledge that some people regard animal experimentation as an unjustified means of pursuing knowledge about human diabetes mellitus. There are those with the strongly held belief that it is morally indefensible for mankind to subject animals to experimental procedures that cause any degree of discomfort, as an animal is unable to provide informed consent and does not benefit directly or indirectly from the studies [2–4]. At the other extreme, a few medical scientists argue that they should be free to perform whatever experiments they consider appropriate with little or no external interference [5]. In the United Kingdom, mainly as a result of public concern regarding vivisection on un-anaesthetized animals, a series of Acts of Parliament have been passed to control the use of animal experimentation (Table 1) [6,7]. It is important to note that legal constraints and the strictness with which they are enforced vary considerably around the world and even within Europe.

Table 1.  United Kingdom Acts of Parliament governing use of animals (a) and key aspects of the Animal (Scientific Procedures) Act 1986 (b)
Martin's Law 1822
Cruelty to Animals Act 1876
Protection of Animals Act 1911
Animals (Scientific Procedures) Act 1986
Protected animals
Any living vertebrate (except man) and the common octopus
Regulated procedures
Experimental or other scientific procedures applied to a protected animal which may have the effect of causing that animal pain, distress, suffering or lasting harm
Personal licences
The personal licence (PIL) authorizes persons with proper competence to actually perform the regulated procedures but only in conjunction with a project licence.
Project licences
The project licence (PPL) authorizes the specified work for a specified purpose with due regard to the balancing of benefit against severity.
Designation of establishment
The work must be performed in an establishment authorized for experimentation and/or breeding and supply. Each establishment must have a named animal care and welfare officer and a named veterinary surgeon.

In the United Kingdom, animal research is currently controlled by the Animals (Scientific Procedures) Act 1986. This defines protected animals as any living vertebrate (except man) and the common octopus (Octopus vulgaris). Any experimental or other scientific procedures applied to a protected animal which may have the effect of causing that animal pain, distress, suffering or lasting harm are termed ‘regulated procedures’ and are illegal unless covered by appropriate licences. Note that killing a protected animal by certain methods (listed in Schedule 1 of the Act) does not require a licence, although appropriate training and skill is mandatory. An individual researcher must hold a Personal Licence (PIL) that authorizes them to actually perform regulated procedures on protected animals. The PIL is only granted after appropriate training and assessment and specifically lists those regulated procedures and protected animals covered by the licence. However, the PIL can only be used in conjunction with a Project Licence (PPL). The PPL authorizes specified work for a specific purpose and must include a detailed protocol for each regulated procedure. Each procedure must be assessed for severity. In addition, the research can only be carried out in an institution that holds a ‘Designation of Establishment’ and this will include the appointment of a named animal care and welfare officer and a named veterinary surgeon. The entire system is overseen by a Home Office Inspector.

Thus, in the UK, a researcher wishing to embark on diabetes research involving animals will have to be working in a designated establishment and apply for both Project and Personal licences. Applications are reviewed by a local animal research ethics committee prior to submission to the Home Office and a timescale of 6–12 months for full authorization is not unusual. Recently, a high-profile group of UK academics have voiced concerns that they are increasingly unable to compete with research groups working in countries with far less stringent controls on animal experimentation [8].

Even if a certain experiment is legal to perform in the United Kingdom, the administrative arrangements and costs may mean that the study can be performed more quickly and cheaply in another country. Furthermore, some diabetes charities that rely heavily on public donations may come under pressure not to fund animal research [9]. For example, some donors may specifically seek reassurances that their money will not be used for animal research.

Animal models of hyperglycaemia

In the 1880s, von Mering was working on the absorption of fat from the intestine when Minkowski suggested he remove the pancreas of a dog. The animal developed polyuria and polydipsia and was found to have diabetes mellitus. Many experiments on rabbits and dogs followed, although history has given a special place to Marjorie, one of the dogs used by Banting and Best in their seminal experiments on the isolation and purification of insulin in the 1920s [10]. Marjorie is probably the most famous experimental animal in history, only to be superseded by Dolly the Sheep in recent years.

One of the most straightforward ways of studying the effects of hyperglycaemia in an animal is to remove the pancreas, either partially or totally. The species of animal used is determined by several factors. In general, the smaller the animal, the more manageable and cheaper the experiment. Hence, pancreatectomised rats and mice are the most commonly used. One of the guiding principles of animal research is to use the ‘lowest’ possible animal and, nowadays, permission would not be granted to remove the pancreas of a dog unless a similar experiment could not be performed in a rodent. However, a major criticism of using rodents is that they may not adequately reflect the human situation and occasionally justification is sought to use larger animals such as cats, dogs, pigs and primates.

Non-surgical methods of inducing hyperglycaemia by damaging the pancreas also exist. These include the administration of toxins such as streptozotocin [11] and alloxan [12]. Table 2 lists a variety of agents known to produce diabetes in experimental animals and streptozotocin is discussed further below.

Table 2.  Substances with a diabetogenic effect in experimental animals

Surgical and toxin-mediated pancreatic damage are valuable tools in the study of the consequences of hyperglycaemia, e.g. the development of diabetic complications [13–19]. When performed on female animals, they may be used to study the effect of gestational diabetes on the offspring [20–23]. Trying to control the hyperglycaemia by the administration of insulin or oral hypoglycaemic agents can further refine the experiments [24]. However, it is almost impossible to restore normoglycaemia in pancreatectomised animals [25].

Much of the research on islet cell transplantation has relied on the use of rodents previously rendered diabetic by the techniques described above. It is then possible to transplant islets and study the consequences. Islets may be transplanted with or without capsules designed to inhibit rejection [26], either subcutaneously [27], under the renal capsule [28] or seeded by the portal vein into the liver [29]. Animals can then be given a variety of anti-rejection therapies [30–32]. Even if human patients were to give informed consent for such studies, it is difficult to envisage how such a large number of experimental conditions could be replicated without using animal models. Nevertheless, it is important to remember that what works in a rat may not work in a human and that the ultimate success of islet cell transplantation programmes will be gauged by data from human studies.

Away from the high-profile field of islet-cell transplantation, diabetic rodents are routinely used in the assessment of new pharmacological agents with a potential role for treating humans with the disease. Thus, all new orally active agents will be tested for efficacy and safety in rodent models. Sometimes this leads to potentially interesting and unexpected findings, such as the possibility that PPARγ agonists may preserve β-cell mass [33]. Other times, less welcome results are obtained, such as the development of tumours in rats treated with an insulin analogue [34].

Animal models of glycosuria

Phlorizin is a naturally occurring flavinoid that, when injected subcutanously, binds to the Na/glucose transporter in the proximal renal tubule and inhibits the resorption of glucose [35]. The administration of phlorizin to a rodent with diabetes can produce near-normoglycaemia [36], at the expense of massive glycosuria and polyuria. This model may be used to study the role of hyperglycaemia in the development of diabetic complications [37–41] and can help disentangle the effects of glucose from the myriad of other metabolic consequences of experimental diabetes.

Animal models of Type 1 diabetes mellitus

Type 1 diabetes mellitus in humans is characterized by a specific destruction of the pancreatic β cells, commonly associated with immune-mediated damage [42]. Although the damage may occur silently over many years, at clinical presentation there is little surviving β cell mass and the disorder progresses to absolute insulinopaenia.

Streptozotocin is a nitrosurea derivative isolated from Streptomyces achromogenes with broad-spectrum antibiotic and anti-neoplastic activity [43]. It is a powerful alkylating agent that has been shown to interfere with glucose transport [44], glucokinase function [45] and induce multiple DNA strand breaks [46]. A single large dose of streptozotocin can produce diabetes in rodents, probably as a result of direct toxic effects. Alternatively, multiple small doses of streptozotocin are used (e.g. 40 mg/kg on five consecutive days). In susceptible rodents this induces an insulinopenic diabetes in which immune destruction plays a role, as in human Type 1 diabetes.

The multiple low-dose streptozotocin model has been used extensively to study the immunological pathways that lead to insulitis and β cell death [47–52]. However, the agent will produce diabetes even in the absence of functional T and B cells [53] and, in contrast to the spontaneous animal models discussed below, diabetes cannot be reliably transferred to syngenic recipients by the transfer of splenocytes [54].

Spontaneous animal models of Type 1 diabetes

The non-obese diabetic (NOD) mouse and bio breeding (BB) rat are the two most commonly used animals that spontaneously develop diseases with similarities to human Type 1 diabetes (Table 3). It should be noted that these animals have been inbred in laboratories for many generations, by selecting for hyperglycaemia. As a result of this process, many genes and phenotypes will have been enriched, but not all will be relevant to the pathophysiology of diabetes, either in rodents or in humans.

Table 3.  Animal models of Type 1 diabetes
NOD (non-obese diabetic) mouse
BB (bio breeding) rat
LETL (Long Evans Tokushima lean) rat
New Zealand white rabbit
Keeshond dog
Chinese hamster
Celebes black ape (Macacca nigra)

The NOD mouse

The NOD mouse was developed by selectively breeding offspring from a laboratory strain that in fact was first used in the study of cataract development (the JcI-ICR mouse) [55]. Insulitis is present when the mice are 4–5 weeks old, followed by subclinical β-cell destruction and decreasing circulating insulin concentrations. Frank diabetes typically presents between 12 and 30 weeks of age. Unlike human Type 1 diabetes, ketoacidosis is relatively mild and affected animals can survive for weeks without the administration of insulin. Also, in contrast to the findings of most studies in humans, there is a larger gender difference with 90% of females, but only 60% of males developing diabetes in some colonies [56].

In comparison with other animal models used in biomedical research, the NOD mouse seems particularly analogous to human Type 1 diabetes and thus a great many studies have been performed in the 25 years since its development. For example, genetic studies have localized multiple susceptibility genes and, as with humans, the MHC region plays a key role [57]. Genetic linkage studies in humans are always hampered by the need to collect a large number of families, concerns regarding paternity, and the reliance on whatever pattern of mating has occurred within the population. Animal studies circumvent all of these problems as it is possible to study large numbers of offspring from whichever mating the researcher chooses to set up.

Inbred animals such as the NOD mouse also have benefits when studying other features of diabetes, as genetic heterogeneity need not be considered as a confounding factor. For example, studies have tried to piece together the immunological cascade that includes T-helper type 2 (Th2) cells, effector cells (CD4+, CD8+) and the contribution of cytokines [58–62].

The BB rat

The BB rat was first recognized in the Bio Breeding Laboratories, a commercial breeding company based in Ottawa, in 1974 [63]. In diabetes prone strains, weight loss, polyuria, polydipsia, hyperglycaemia and insulinopenia develop at around 12 weeks of age, often at the time of puberty. In common with the human disease, ketoacidosis is severe and fatal unless exogenous insulin is administered [64]. As with the NOD mouse, the pancreatic islets are subjected to an immune attack with T cells, B cells, macrophages and natural killer cells being recruited to the insulitis [65–69]. A variety of auto-antibodies, including GAD, have been reported in both BB rats and the NOD mouse, although it remains far from clear which, if any, of these are primary autoantigens [70,71].

BB rat strains prone to diabetes typically have profound T-cell lymphopenia in the circulating blood, specifically lacking T cells that express ART2 [72]. Transfusion of histocompatible T cells expressing ART2 will prevent the spontaneous development of hyperglycaemia in BB rats. Genetic studies have identified several susceptibility genes, including loci in the MHC region [73–75].

Prevention of diabetes in NOD mouse and BB rat

Rodent models of Type 1 diabetes have been employed in research examining the role of diet (e.g. cow's milk proteins) [76–79] and a variety of viruses [80–84] as environmental triggers for the disease. For example, infection with the mouse hepatitis virus decreases the incidence of diabetes in NOD mice, whilst infection with Kilham rat virus increases the risk in BB rats [85]. Immunosupression with cyclosporin in combination with other immuno-modulatory compounds such as vitamin D helps prevent diabetes in both BB rats and NOD mice [86,87]. Non-specific immunization with Freund's adjuvant will also prevent diabetes in these rodent models [88,89]. Insulin has been administered orally in the hope of inducing tolerance and reducing the immune attack on the pancreatic islets. Some, but not all, studies have shown this strategy can prevent diabetes in the NOD mouse and BB rat [90,91]. Nicotinamide is known to protect β cells from toxins such as alloxan and also helps prevent the onset of diabetes in NOD mice [92].

As a result of these observations, several trials in humans at high risk of diabetes have been undertaken, including the use of oral insulin and nicotinamide. Unfortunately the results are currently disappointing [93].

Animal models of Type 2 diabetes

Type 2 diabetes represents a heterogeneous group of disorders characterized by insulin resistance and impaired insulin secretion and defined by a raised fasting or post-challenge blood glucose. Some subtypes of diabetes are now recognized as being because of specific single gene defects [e.g. the maturity-onset diabetes of the young (MODY) syndromes [94], syndromes of severe insulin resistance [95] and mitochondrial diabetes [96]]. However, for most patients with diabetes, several (if not many) genetic and environmental factors contribute to the causation and progression of the disease and also the late complications.

Animal models of Type 2 diabetes are likely to be as complex and heterogeneous as the human condition. Advances are most likely to come from interpreting data from several sources (Table 4). Thus, in some animals, insulin resistance predominates, whilst in others β-cell failure is pre-eminent. Models where glucose intolerance is part of a wider phenotype of obesity, dyslipidaemia and hypertension may also provide valuable insights into human Type 2 diabetes. As with the NOD mouse and BB rat in Type 1 diabetes, the selective inbreeding of animals that spontaneously develop a Type 2 diabetes-like phenotype has generated many of the strains used today. Much can also be learnt from animals with single gene mutations, as evidenced to by the advances in knowledge generated from the study of the ob/ob, db/db, fa/fa and agouti strains [97–101].

Table 4.  Animal models of Type 2 diabetes mellitus
Ob/Ob mouse—monogenic model of obesity (leptin deficient)
db/db mouse—monogenic model of obesity (leptin resistant)
Zucker (fa/fa) rat—monogenic model of obesity (leptin resistant)
Goto Kakizaki rat
KK mouse
NSY mouse
Israeli sand rat
Fat-fed streptozotocin-treated rat
CBA/Ca mouse
Diabetic Torri rat
New Zealand obese mouse

Rodent models of monogenic obesity and diabetes

Obesity and the consequent insulin resistance is a major harbinger of Type 2 diabetes mellitus in humans. Consequently, animal models of obesity have been used in an attempt to gain insights into the human condition. Some strains maintain euglycaemia by mounting a robust and persistent compensatory β-cell response, matching the insulin resistance with hyperinsulinaemia. The ob/ob mouse and fa/fa rats are good examples of this phenomenon. Others, such as the db/db mouse and Psammomys obesus (discussed later) rapidly develop hyperglycaemia as their β-cells are unable to maintain the high levels of insulin secretion required throughout life. Investigation of these different animal models may help explain why some humans with morbid obesity never develop Type 2 diabetes whilst others become hyperglycaemic at relatively modest levels of insulin resistance and obesity.

Significant advances have been made in the field of obesity research using these rodent models. In 1994, Friedman and colleagues reported the cloning of the mutation responsible for the massive obesity seen in the ob/ob mouse [99]. The defective ob gene coded for a protein secreted by adipocytes and termed leptin. The db/db mouse and fa/fa rat were subsequently found to harbour mutations in a hypothalamic receptor for leptin [100,101].

The Goto Kakizaki (GK) rat

The GK rat was developed by the selective breeding of Wistar rats with the highest blood glucose over many generations [102]. The rats develop relatively stable hyperglycaemia in adult life. Typically, the fasting blood glucose is only mildly elevated but rises further on challenge with glucose. Both insulin resistance and impaired insulin secretion are present. At birth, the GK rat has a reduced number of islets [103]. Inheritance is polygenic and several genome-wide scans have identified putative susceptibility loci on a variety of chromosomes [104–106]. In common with human Type 2 diabetes mellitus, an excess of maternal transmission has been reported in some but not all studies [107].

The GK rat, in common with other animal models of diabetes, develops some features that can be compared with the complications of diabetes seen in humans. These include renal lesions [108], structural changes in peripheral nerves [109] and abnormalities of the retina [110]. Research into these phenomena represents another example of how animal experimentation opens up areas that could not be easily studied in man. For example, serial renal or nerve biopsies cannot be ethically performed in human patients and it is similarly difficult to measure the concentrations of intraocular growth factors without recourse to rodent models. However, there are certainly differences between the complications seen in animal models and humans and debate exists as to whether any currently available animal model accurately reflects the diabetic complications seen in man.

The KK mouse

The original strain of this mouse was bred for large body size [111]. Characteristically, the mouse gradually becomes obese in adult life, associated with insulin resistance, compensatory hyperinsulinaemia and islet cell hyperplasia. Eventually, mild hyperglycaemia supervenes [112]. Food intake is important in determining the severity of the diabetic phenotype and restriction of energy intake reduces both the obesity and hyperglycaemia seen in this strain of mice. Several different lines have been bred throughout the world and, because of selective breeding and founder effects, these vary both genetically and phenotypically from each other. For example, in one line of KK mice, a glycoprotein gene (Azgp1) has been postulated as playing a role in susceptibility to obesity and diabetes [113], but it is not clear whether this is true for other strains. Although all KK mouse strains can ultimately be traced back to animals distributed from the original colony described in 1967 [111], some differences may now exist in colonies maintained in different laboratories throughout the world. Reasons for this include the difficulty in ensuring 100% genetic homogeneity amongst inbred strains, even after many generations of selective breeding. Furthermore, new spontaneous mutations may occur in isolated breeding colonies in different laboratories and these can become ‘fixed’ in one particular colony and not another. This might explain the lack of concordance between studies designed to identify the precise genetic susceptibility loci in this animal model of diabetes [114]. Such differences should always be borne in mind when assessing results from similar experiments performed by different laboratories that are apparently using the same animal model of diabetes.

The decision of which rodent model of diabetes to use for any particular experiment is often multifactorial. Ideally, experiments should be carried out in several different types of animal although, in practice, individual research groups tend to build up experience with one strain. For example, reasons for using the KK mouse rather than the GK rat include the superiority of the KK mouse in mimicking human obesity (the GK rat being relatively slim) and the generally easier production of transgenic variants from mice rather than rats. It is important to realize that, in general, one animal model represents only one aspect or subtype of diabetes in humans and care must always be taken when extrapolating results to the clinical setting

The Nagoya–Shibata–Yasuda (NSY) mouse

As with the KK mouse, this model was developed by selective inbreeding, but on this occasion using a laboratory strain of mouse termed Jc1:ICR (which is also the ancestral strain from which NOD mice were developed). NSY mice spontaneously develop diabetes in an age-dependent manner. Key features include impaired insulin secretion in the face of mild insulin resistance. Obesity is not a major feature of these animals and there is a marked gender difference with almost all males developing hyperglycaemia, but less than a third of females, being affected [115]. This gender difference is a generalized characteristic for most animal models of Type 2 diabetes mellitus.

The NSY mouse is particularly useful when considering age-related phenotypes (e.g. decline in β-cell function). Of interest, a genome-wide linkage scan has identified a sequence variation in the hepatic nuclear factor 1β gene, a gene also implicated in human MODY syndrome [116].

Psammomys obesus (the Israeli sand rat)

In its natural habitat, the Israeli sand rat has an essentially vegetarian diet. However, when fed laboratory chow, the animals become obese, insulin resistant and hyperglycaemic [117]. If a cholesterol-rich diet is used, hyperlipidaemia and atherosclerosis develop [118]. In common with human Type 2 diabetes mellitus, the hyperglycaemic state is associated with an increase in circulating proinsulin and split products presumably because of the high demand for insulin secretion driven by insulin resistance. Impaired insulin biosynthesis within the islets has also been reported [119].

As with the KK mouse, the Israeli sand rat model is particularly useful when studying the effects of diet and exercise [120] on the development of Type 2 diabetes.

The Otsuka Long-Evans Tokushima fatty (OLETF) rat

The OLETF rat originates from an outbred colony of Long-Evans rats selectively bred for glucose intolerance [121]. The rats are mildly obese and, as with NSY mice, males are more likely to develop diabetes in adult life than females. Genome-wide scans have reported susceptibility loci on chromosomes 1,7,14 and also the X chromosome [122,123]. Interestingly, OLETF rats also carry a null allele for the cholecystokinin A receptor which may be involved in the regulation of food intake [124], however, whether this is causally related to the phenotype is still unclear. Genetic studies in both animals and humans are complex, given the many susceptibility and protective loci that contribute to the overall risk of diabetes and/or obesity. Animal studies have the advantage of allowing the study of large numbers of individuals with precisely known ancestry and the ability to set up back-crosses and matings at will. However, inbreeding will also enrich genetic variants in a strain that may have no causal relationship to diabetes itself.

Gene targeting and transgenic techniques

An explosion in the number of animal models used for the study of diabetes has come from the use of molecular biological techniques. Gene targeting refers to the process by which a single gene may be disrupted in an embryonic stem cell and then transmitted along the germ cell line. This produces ‘knockout’ animals. Transgenics refers to the incorporation of modified genes (transgenes) into the pronucleus of a zygote. The transgene is randomly incorporated into the host genome and some offspring will therefore express the modified gene. Theoretically, these techniques allow the selective perturbation of a single element of a complex system. For example, rodents may be produced that over- or under-express proteins thought to play a key part in glucose metabolism.

Briefly, it is possible to superovulate a female mouse or rat and allow her to mate. The following day, the single cell zygotes are harvested and maintained in culture for a few hours. A genetic construct (e.g. the insulin gene) can then be injected into the pronucleus of the zygote. The construct will integrate itself into the genome of the zygote and, if this occurs at an appropriate part of the host DNA, the result is an embryo that will over-express insulin in adult life. By including specific promoter regions in the genetic construct, it is possible to gain some control over how the transgene is expressed in adult life. For example, by including the metallothionein promoter, the inserted gene will be ‘switched on’ by the in-vivo administration of heavy metals to the adult (Fig. 1).

Figure 1.

Producing transgenic animals that over-express a gene.

A number of factors must be taken into account when studying the offspring of these transgenic experiments. As the transgenes are incorporated at random into the host genome, the effect observed in offspring will vary depending on precisely where the integration has occurred and also the number of copies that have successfully integrated. In some cases the transgene disrupts a native gene at the site of its incorporation into the genome, further complicating the interpretation of the resultant phenotype. In addition, the effect of transgene expression is strongly modified by the genetic background of the mice, so that results may vary depending on which particular strain is employed.

Knockout animals are produced by using a genetic construct that will disrupt a normal gene. Typically, a construct is developed that contains DNA sequences homologous to the target gene but that are disrupted or contain a deletion. These are injected into embryonic stem (ES) cells and will undergo recombination with the normal gene, causing it to be ‘knocked out’. The construct is usually engineered to contain an antibiotic resistance sequence and this allows selection of the ES cells that have successfully incorporated the DNA. ES cells are then injected into pre-implantation mouse embryos and transferred to the oviducts of pseudopregnant mice and allowed to develop to term. When the ES cells have contributed to the germ line of the offspring, selective breeding will allow the production of mice that are either heterozygous or homozygous for the knockout (Fig. 2).

Figure 2.

Steps in the production of knockout animals.

Using these approaches, a large number of animals have been produced in an attempt to gain insights into the pathogenesis of both Type 1 and Type 2 diabetes. Some of these are listed in Table 5.

Table 5.  Transgenic animals in diabetes research
GeneCopy numberPhenotypeReference
Insulin receptor0Severe hyperglycaemia, neonatal death from ketoacidosis125
Insulin receptor1 (in muscle)Mild insulin resistance126
IRS10Mild insulin resistance 
Growth retardation 
Insulin resistance128, 129
Reduced β-cell mass 
Glucokinase0Severe hyperglycaemia
Perinatal death
Glucokinase1Non-progressive glucose intolerance130
GLUT4> 2Increased insulin sensitivity131

The targeted gene disruption and transgenic approaches are not without limitations. Simply over- or under-expressing a single gene will result in many others being altered as a compensatory mechanism. For example, completely knocking-out the insulin receptor substrate 1 (IRS1) gene produces remarkably little in the way of effect given the central role that this protein has in intracellular signalling [127]; redundancy within the insulin signalling cascade overcomes the genetic manipulation. Homozygote knockouts of the insulin receptor substrate 2 (IRS2) gene, however, do develop diabetes by 10 days of age, associated with insulin resistance in the liver and reduced pancreatic β-cell mass [128].

Other difficulties with classical gene disruption techniques include the fact that homozygote knockouts are sometimes lethal in utero so adults cannot be studied. Some genes have quite different functions during embryogenesis than in adult life and the knockout cannot be timed to occur during specific periods of the animal's life. In addition, animals will have altered gene expression in many tissues, not necessarily just those of physiological importance. One way of overcoming this limitation is to produce animals in which the gene being studied is only knocked out in specific tissues.

Tissue-specific knockouts

Cre is a bacteriophage P1 recombinase enzyme that recognizes specific sequences of DNA 34-bp long (LoxP sites). When two of these sites occur close together, Cre will cut out the DNA between them. In the laboratory, it is possible to engineer a DNA construct in which Cre is placed next to a tissue-specific promoter (e.g. the insulin gene promoter). Using the techniques for creating transgenic animals outlined above, offspring can be produced that will have the Cre construct in all their tissues, but it will only be active in sites where the insulin gene promoter is active (i.e. pancreatic β cells). A second set of transgenic animals can be produced that have, for example, the insulin receptor gene flanked by the target lox P sites (Fig. 3). By mating the two animals together, one can develop a line of animals where the insulin receptor gene is only knocked out in pancreatic β cells. Similarly, animals can be produced that have a gene knocked out in other specific tissues such as liver, adipose tissue or brain.

Figure 3.

Production of tissue-specific knockout animals.

This very powerful technique has produced much interesting data in recent years. For example, an animal model of mitochondrial diabetes was developed by knocking out the mitochondrial transcription factor mtFA only in the pancreatic β cells [132]. If mtFA had been knocked out in all tissues of the animal, as occurs in classical transgenics, the phenotype would have been very difficult to interpret and it would not have been possible to answer specific questions about the role of mtDNA defects within β cells. The adult animals developed insulinopenic diabetes, similar to diabetes as a result of mtDNA defects in humans. Some other tissue-specific knockouts that have been developed in recent years are shown in Table 6.

Table 6.  Tissue-specific knockouts in diabetes
Genetic knockoutTissue specificityReference
Insulin receptor β  cells, brown fat, liver, muscle, brain133
PPARγ β  cells134

Animal models of diabetes in pregnancy and the role of intrauterine environment

Another important field of diabetes research that has relied heavily on animal experimentation is the study of diabetes in pregnancy and the role of the intrauterine environment on the subsequent development of diabetes amongst offspring.

Firstly, it has been shown that diabetes in pregnancy predisposes to the later development of diabetes amongst offspring. Thus, pups born from rats rendered diabetic by streptozotocin treatment are more likely to become diabetic in adult life than litters born from the same mother before she received the streptozotocin [138]. In embryo transfer experiments, Wistar rats (at low genetic risk of diabetes) are more likely to develop hyperglycaemia as adults if they are reared in the uterus of a Goto Kakizaki (diabetic) mother than a euglycaemic mother. However, transferring Goto Kakizaki embryos into a normal (Wistar) uterus does not seem to reduce their risk of developing diabetes [139].

Secondly, it has been shown that intrauterine malnutrition may also increase the risk of diabetes amongst offspring in later life. This has been achieved by a variety of means, including uterine artery ligation [140] and dietary restriction of pregnant dams [141]. Endocrine pancreas development is altered in foetuses from rats previously showing intrauterine growth retardation in response to malnutrition [142].

The diabetogenic effects of manipulating the intrauterine environment are probably mediated by a permanent programming of the developing offspring, e.g. by the mechanism of imprinting. Of interest, the increased risk of diabetes continues into subsequent generations, suggesting the changes also affect the germ cell line [143].


There is little doubt that some animal models of diabetes have provided an invaluable insight into the pathogenesis of the human disease. Patients have directly benefited from the use of animals in the discovery of insulin and the assessment of other treatments. However, there have been ‘blind-alleys’ in research as well, such as the lack of reproducible paradigms of human diabetic complications and the disappointing results of studies aimed at preventing Type 1 diabetes based on strategies that are successful in rodents.

Animal research in the United Kingdom is governed by the three Rs—reduction, refinement and replacement. Strenuous efforts should be made to reduce the number of animals required for any particular study, experimental design should be refined so that it causes the least discomfort to the animals and, ultimately, animal models should be replaced altogether. Care should always be taken when comparing results from apparently similar experiments performed in different institutions and when extrapolating results from animal experiments to the human situation. Despite these caveats, significant advances in knowledge have arisen in recent years, especially with the advent of transgenics. It is likely that animal models will play an important role in the eventual cure of human diabetes mellitus.

Competing interests

None declared.