A Review of Islet of Langerhans Degeneration in Rodent Models of Type 2 Diabetes

Abstract

Type 2 diabetes mellitus (TTDM) is characterized by progressive loss of glucose control through multifactorial mechanisms. The search for an understanding of TTDM has relied on animal models since the realization of the importance of the pancreas in controlling plasma glucose concentration. Rodent models of TTDM are developed to express hyperglycemia and not islet degeneration per se. Degeneration of the islets of Langerhans with β-cell loss is secondary to insulin resistance and is regarded as the more important lesion. Despite this, differences between models are seen in the development and progression of islet degeneration. Assessing the differences between the models is important to appreciate the various aspects of TTDM and understand their advantages as well as their deficiencies. Relevant animal models of TTDM provide opportunities to investigate important physiological and cell biological processes that may ultimately lead to development of targeted therapies. This article reviews the importance, advantages, and limitations of rodent models of TTDM in relation to the histopathological changes that characterize islet degeneration. Pathophysiological mechanisms that contribute to islet degeneration are also discussed and are placed into the context of changes in islet histological appearances.

Introduction

Type 2 Diabetes Mellitus in Human

Diabetes mellitus is prolonged fasting hyperglycemia due to a deficiency in insulin signaling. Type 2 diabetes mellitus (TTDM) has been described as a global pandemic with an estimated incidence of 151 million in the year 2000 (Engelgau et al., 2003). Of these, approximately 90–95% presented with TTDM and the remainder with type 1 diabetes mellitus (TODM). For most of the 20th century, TTDM was most frequently diagnosed in people older than age 50, but more recently has been found increasingly in younger individuals (Drake et al., 2002). As a consequence, estimates of the prevalence of TTDM in the United States are expected to double from 2000 to 2025 (Barceló and Rajpathak, 2001), and the estimated global incidence to exceed half a billion by 2010 (Hansen, 1999).

Diabetes mellitus (types 1 and 2) both involve a reduction in plasma insulin concentrations. In TODM, insulin signaling is completely annulled due to absolute insulinopenia caused by autoimmunity directed against, and which ultimately results in the destruction of, the insulin producing cells (β cells). Since TODM is the more severe form of diabetes, the affected patients are insulin dependent and susceptible to ketoacidosis, which is a feature of only the most severe cases of TTDM (Otieno et al., 2005). In contrast to TODM, patients presenting with TTDM do not have islet-autoreactive antibodies (insulitis), but show a more gradual loss of insulin signaling through multifactorial mechanisms including insulin resistance and β-cell dysfunction. Glucose intolerance, followed by fasting hyperglycemia are the most widely recognized and important clinical features of these defects (Abdul-Ghani et al., 2006). Insulin resistance precedes TTDM, and insulin resistance syndrome has been recognized as an important risk factor for TTDM and cardiovascular disease (Pfützner et al., 2004). Hyperinsulinemia maintains normoglycemia in insulin resistant individuals for a protracted period of time before the development of TTDM, which is preceded by a β-cell defect resulting in reduced insulin output. Dysfunction in β-cell insulin production, glucose sensing, and cell turnover are among the many defects that have been correlated with the degeneration of islets of Langerhans in models of TTDM.

The Pathogenesis of TTDM in Humans

Insulin resistance syndrome is one of several terms (including syndrome X and metabolic syndrome) that are used to describe a collection of clinical conditions that pertain to cardiovascular disease risk factors and is considered to be a prediabetic syndrome (Zimmet et al., 2001). Insulin resistance is characterized by a desensitization to insulin-dependent processes, including glucose uptake in insulin-sensitive tissue, glycogenesis, suppression of adipose tissue lipogenesis and hepatic glucose output (Reaven, 1995). Insulin resistance has been strongly associated with obesity and is age-dependent, becoming more severe with age (Perry et al., 1995). In some cases, specific genetic causes, such as type A insulin resistance, acanthosis nigricans, and congenital insulin resistance, have been recognized (Moller and O’Rahilly, 1993). During the course of the syndrome, β cells initiate a compensatory response and hypersecrete insulin, with resulting hyperinsulinemia. Although insulin resistance precedes overt β-cell failure, both insulin resistance (Mari et al., 2005) and alterations in β-cell-phased secretion patterns are seen in nondiabetic, first-degree relatives of TTDM patients (Gerich, 2003; Vauhkonen et al., 1998). These alterations in β-cell activity include significantly reduced insulin secretory response to a glucose challenge and reduced compensation for insulin resistance (Mari et al., 2005).

As the β-cell output of insulin becomes insufficient to maintain normoglycemia, a decrease in insulin output marks the onset of β-cell failure and TTDM (Kahn, 2003). TTDM patients often show plasma insulin levels equal to lean nondiabetics, although this level is reduced in comparison with obese, weight-matched, insulin-resistant individuals. Identification of the primary defect, relative contributions, and natural order of insulin resistance and β-cell dysfunction has proved difficult and varies between patients (Ferrannini, 1998), but insulin resistance alone does not cause diabetes. Indeed, studies have shown a lower frequency of diabetes development in a high-risk (prediabetic) population when treated with rosiglitazone, a sensitizer to insulin through peroxisome proliferator-activated receptor-gamma (PPAR-gamma) agonism, indicating that the progression of diabetogenesis is not absolute and may be controlled before overt TTDM is established (Gerstein et al., 2006).

Despite gaps in our knowledge, it is known that diabetes does not occur without progressive β-cell dysfunction. The purpose of this article is to discuss the mechanisms of β-cell dysfunction, their progression, and how these are manifested histopathologically as degeneration in the islets of Langerhans in humans and rodent models of the disease.

Islets of Langerhans

Islets constitute about 2% by weight of the adult human pancreas and are multicellular microorgans (Bonner-Weir, 2005; Klöppel and Veld, 1997). In descending order of both cell number and cell mass, islet cell types are β (insulin secreting, 70–80%), α(glucagon secreting, 15–20%), PP (pancreatic poly-peptide secreting, 15–20%), δ(somatostatin secreting, 5–10%), and the recently discovered ghrelin-secreting cells (ghrelin, 1%; Wierup et al., 2002). Normal islet architecture differs between species and between regions of the pancreas (Figure 1). Immunohistochemical staining of normal adult human islets with anti-insulin antibodies show a diffuse mass of β cells with other islet cell types distributed among them. In contrast, islets of nondiabetic strains of rodent show a large β-cell core, with other islet cells forming an enveloping mantle that is sometimes continuous but often is discontinuous.

The islet is an important controlling unit of metabolism and the endocrine cellular mass is dynamic and reactive to changes in secretory demand. Systemic metabolic changes due to insulin insensitivity and loss of glucose control are mirrored by changes in islet structure. The islet increases (hypertrophy) and decreases (atrophy) in size and function with worsening diabetic state. Islet hypertrophy and insulin hypersecretion are stimulated by an increased requirement for insulin, which may be due to hyperphagia, insulin resistance, or, most commonly, a combination of the two. Ogilvie (1933) first observed human islet hypertrophy associated with obesity.

Islet Histopathology in Humans

Limited information is available on the progressive degeneration in β-cell function during the development of TTDM in humans. End-stage diabetes is characterized by a reduction in β-cell mass, the deposition of intra-islet amyloid (IIA), and by fat deposition into the islets. Postmortem histopathological investigations are normally performed only on individuals following long-standing diabetes, and as noninvasive techniques for visualizing islets are presently unavailable, little is known about changes in human islet histopathology during early diabetes. However, islet hypertrophy has been reported in obese patients (Ogilvie, 1933) and amyloid deposits seen in a minority of elderly patients that had not previously presented with TTDM (Zhao et al., 2003). These are high-risk groups for the development of TTDM, and it is therefore possible that such islet changes reflect prediabetic changes.

Human Pancreatic Amyloidosis

Pancreatic amyloid deposits were first described by Opie (1900) as intracellular (β) and extracellular hyaline changes. Human islet amyloid polypeptide (hIAPP) is a soluble, 37-amino-acid peptide that is cosecreted with insulin and forms from cleavage of the 67-amino-acid pro-IAPP (Johnson et al., 1991). Amyloid plaques form when conformational changes in hIAPP cause it to become insoluble and fibrillar. This is characteristic of TTDM pancreatic histopathology and Clark et al. (1987, 1995) proposed that hIAPP deposits are pathogenic in TTDM. A biochemical change preceding this process has not been identified and amyloid formation may simply be due to a physical concentration of hIAPP in the extracellular space. In healthy humans, IAPP is present in plasma at 5–15 pM and Wang et al. (1997) have suggested that hIAPP inhibits the secretion of insulin and induces the feeling of satiation (Isaksson et al., 2005). Pancreatic amyloid formation occurs in diabetic humans, in non-human primates (de Koning et al., 1993), and in cats (Yano et al., 1981), but has not been reported in rodents. Amyloidogenicity is prevented in rodents by proline substitutions at amino acids 24–29 of the molecule. This difference is thought to be due to hydrophobic residues on hIAPP that cause the molecule to form insoluble fibrillar amyloid at high concentrations (Westermark et al., 1990).

The role of amyloid in the pathogenesis of TTDM is unclear. The largest study to date showed pancreatic amyloidosis in 36.2% of 235 TTDM versus 3.0% in 533 nondiabetic Chinese patients (Zhao et al., 2003). Furthermore, pancreatic amyloid formation shows a substantial correlation with high Hb_A1C (a marker of long-term hyperglycemia) and other pancreatic pathologies such as islet fibrosis, arteriosclerosis, and parenchymal fat infiltration. Data have also shown a link between the severity of TTDM and amyloidosis, because patients that required exogenous insulin therapy tended to show more severe amyloidosis (Höppener et al., 2000). In addition, patients with end-stage diabetic renal disease showed reduced clearance of circulating IAPP, leading to a higher incidence of systemic amyloid deposits (de Koning et al., 1995). Other studies have found higher incidences of amyloidosis in TTDM patients (92% in Swedish patients: Westermark et al., 1997; 77% in Pima Indians: Clark et al., 1990; 90% in Caucasians: Clark et al., 1996). The comparatively low incidence of amyloidosis found in Chinese patients may have been due to differences in diabetic pathophysiology between racial phenotypes. A much lower incidence of obesity (body mass index [BMI] of > 30) is present in a Chinese, compared with a Caucasian population: A BMI above 23 is considered a risk factor for metabolic disorder in Chinese patients (Weng et al., 2006). Although the study of Zhao et al. (2003) carried higher statistical significance than other, similar studies, the ostensibly low incidence of amyloidosis may illustrate that this is associated with a Western, rather than a Chinese, diabetic pathophysiology. Indeed, it would be informative to study the racial correlation between BMI and pancreatic amyloidosis.

Impaired Proinsulin Cleavage

Increases in plasma proinsulin (intact proinsulin and 32/33 split proinsulin, Clarke et al., 1992) are a consistent finding in human TTDM (Gottsater et al., 1996) and in rodent models of TTDM. Increased insulin secretory demand, rather than cleavage defects, is implicated as the initial, reversible cause of high plasma-proinsulin levels (Alacron et al., 1995). Shunting of insulin granules for release during hypersecretory responses has been shown to result in the release of a greater proportion of immature granules. Proinsulin is ineffective in stimulating insulin-associated pathways, such as glucose uptake, and is also considered an independent risk factor for cardiovascular disease (Pfützner et al., 2004). Hyperproinsulinemia is regarded primarily as a sign of β-cell dysfunction and, importantly, it has been identified as a sensitive marker for insulin resistance (Langenfeld et al., 2004).

Rodent Models of ttdm: Models with Spontaneously Mutated Genes

See Table 1 for characteristics of models described.

Goto Kakizaki Rats

The Goto-Kakizaki rat is a nonobese, diabetic strain that presents with primary β-cell defects and peripheral insulin resistance in both males and females (Bisbis et al., 1993). Goto-Kakizaki rats were Wistar-derived and inbred to conserve glucose intolerance (Östenson, 2001). The colony originated in Japan in 1975 and other colonies have since been established elsewhere, most notably in Stockholm and Paris. At birth, Goto-Kakizaki rats show a paucity of β cells (Movassat et al., 1995, 1997), which may be due to a wave of β-cell apoptosis during embryonic days 16–18 (Miralles and Portha, 2001). Despite this, Goto-Kakizaki rats were normoglycemic and showed normal islet morphology at birth and changes in islets do not develop simultaneously with hyperglycemia. At 4 weeks of age, Goto-Kakizaki rats displayed a mild fasting (hyper) glycemia of 9 mM, raised basal insulin secretion, β-cell insensitivity to glucose, and reduced glucose tolerance (Östenson et al., 1993). This phenotype was found to be resistant to dietary restriction (Alvarez et al., 2000).

By 8 weeks of age, hyperglycemia was marked and islet morphology normal, although slight periislet fibrosis was occasionally seen (Guenifi et al., 1995; Movassat et al., 1997). As the animals age, fibrosis becomes more pronounced and involves intra-islet tissue resulting in islets that display a characteristic “starfish shaped” appearance. Sieca et al. (2003) observed periodic acid Schiff (PAS) positivity in starfish-shaped islets. Furthermore, Homo-Delarche et al. (2006) demonstrated the expression of inflammatory markers, such as MHC class II molecules and CD68, within these islets. Starfish-shaped islets showed increased expression of insulin-like growth factor 2 (IGF-2; Hoog et al., 1996) localized to the membrane of β-cell secretory granules and, to a much lesser extent, in β-cell cytoplasm and in adjacent fibroblasts (Hoog et al., 1997). β-Cell proliferative activity was decreased in adult Goto-Kakizaki rats (as compared with Wistar rats), although there was no difference in neonatal, that is, prediabetic, Goto-Kakizaki rat islets, which suggests that more β-cell defects are present in mature animals than at birth (Portha et al., 2001).

Higher basal rates of islet blood flow and blood pressure have been recorded in the Goto-Kakizaki rat as compared with the Wistar rat strain (Atef et al., 1994; Svensson et al., 1994). During a glucose challenge, however, islet blood flow did not increase as much as that seen in control animals (Carlsson et al., 1997). With age, islet blood flow progressively increases in Wistars, coinciding with a normal, mild desensitization to insulin by 1 year of age. In contrast, Goto-Kakizaki rats do not show the same compensatory capacity to increase islet blood flow, which possibly contributes directly to a deterioration of the diabetic state (Svensson et al., 2000).

Compared with cultured islets obtained from nondiabetic Wistar rats, those from Goto-Kakizaki rats maintained in a medium containing low glucose concentrations show increased glucose-stimulated insulin secretion. However, at high glucose concentrations, insulin secretion is inappropriately low, with an increased response to nonglucose, insulin secretagogues, such as arginine (Abdel-Halim et al., 1993; Hughes et al., 1994). These islets also showed reduced islet insulin content by comparison with age-matched Wistar controls (Movassat et al., 1997; Suzuki et al., 1997). It is noteworthy that when exogenous insulin was used to control glycemia, endogenous insulin was reduced further, indicating that the islets were incapable of being functionally restored.

A defect in glucose metabolism is likely to be the most important factor causing inappropriate glucose-stimulated insulin secretion responses in vitro (Giroix, et al., 1993a, 1993b; Hughes et al., 1998; Nadi and Malaisse 2000; Zaitsev et al., 1997). Several critical defects in the activity of enzymes that control oxidative glycolysis have been identified in Goto-Kakizaki rats, namely, FAD-linked (flavin adenine dinucleotide-linked) mitochondrial glycerophosphate dehydrogenase (mGPD) activity and content (Fabregat et al., 1996; Malaisse, 1993), glucokinase activity (Matsuoka et al., 1995), and pyruvate dehydrogenase activity (Zhou et al., 1995). Changes in other β-cell functions have also been recorded, such as expression of CD38 (Matsuoka et al., 1995) and GLUT2 (Ohneda et al., 1993) were reduced in Goto-Kakizaki islets. However, reduced mGPD and pyruvate carboxylase were normalized when glucose is controlled by exogenous insulin, indicating that defects in these were secondary to diabetes rather than causative factors (MacDonald et al., 1996a). Overexpression of mGDP in Goto-Kakizaki islets has no restorative effect, and therefore supports this notion, whereas in Wistar islets it greatly increases glucose-stimulated insulin secretion (GSIS) at high glucose concentrations (Ueda et al., 1998).

These defects indicate that in β cells obtained from Goto-Kakizaki rats, β-cell glycolysis is inefficient, resulting in a lower production of ATP per molecule of glucose metabolized. This reduction in ATP incurs a reduced response from ATP-sensitive K⁺ channels (K_ATP channels) and therefore results in a reduction in insulin exocytosis. Desensitization of the K_ATP channel to ATP has been reported (Tsuura et al., 1994), but is not believed to be a major cause of reduced insulin secretion. Indeed, studies have shown a hyperresponsiveness to Ca²⁺ in Goto-Kakizaki rat electropermeabilized β cells (Katayama et al., 1995; Okamoto et al., 1995), which may represent a compensatory adaptation to decreased ATP production.

Zucker Fatty and Zucker Diabetic Fatty Rats

The Zucker fatty (ZF) rat has a mutation in the gene coding the leptin receptor (fa/fa) that results in obesity and hypertension with associated renal and cardiovascular disease (Mine et al., 2002). The Zucker diabetic fatty (ZDF) rat also carries the same mutation and, additionally, a mutation that results in spontaneous hyperglycemia at about 7 to 10 weeks of age in males (Peterson et al., 1990; Unger, 1997). Female ZDF rats became hyperglycemic only when fed a diabetogenic diet (Coresttis et al., 1999). Although this additional mutation has not been identified, it is most likely to be expressed in the β cell.

The ZF rat was derived from Wistar rats and the ZDF rat from ZF rats, which were inbred to conserve the phenotypic features of the prediabetic state, such as glucose intolerance (Peterson et al., 1990). Rats that express the fa/– genotype are lean and do not develop hyperglycemia (Wang et al., 1998). The fa/fa mutation of the leptin receptor results in insulin resistance with reduced glucose tolerance. Leptin is a hormone produced predominantly in white adipose tissue and to a lesser extent in brown adipose tissue. Leptin release is stimulated by insulin and acts to suppress insulin secretion (the adipoinsular feedback axis) while inducing satiation (Alemzadeh and Tushaus, 2004). ZF and ZDF rats are hyperphagic due to the reduction in leptin signal that results in obesity. The leptin receptor is found in hypothalamic nuclei, where the endocrine action of leptin differs according to the specific hypothalamic nucleus within which it interacts. Leptin insensitivity and obesity contribute to hyperleptinemia. However, the hyperphagic nature of ZF rats leads to obesity even when caloric intake is controlled, indicating that leptin control of satiation is not the only contributory factor to obesity in this strain. ZDF rats fed a reduced calorific diet showed some restoration of glucose control, but this was insufficient to prevent hyperglycemia (Ohneda et al., 1995).

Adult ZF rat islets show changes consistent with elevated β-cell activity and a prediabetic physiological state (Figures 1 and 2). Some islets appear normal, although many are hypertrophic and a few show a mononuclear cell infiltration, β-cell degeneration and fibrosis. β Cells frequently exhibit reduced insulin granulation. In male ZDF rats, a significant change in plasma concentration of insulin and lipids precedes the onset of hyperglycemia. Initially, the ZDF rat presents with increasing insulin resistance with compensatory hyperinsulinemia that quickly progresses to severe glucose intolerance. Plasma free fatty acids rise rapidly at about 5–8 weeks (Lee et al., 1997) and hyperglycemia occurs between 7 and 10 weeks. The animals become severely dyslipidemic, with raised plasma free fatty acids, cholesterol, and triglyceride. Adult ZDF rats also develop complications to the diabetic state, including cardiovascular disease, peripheral neuropathy (Shimoshige et al., 2000), and renal disease (Erdely et al., 2004).

Before the onset of hyperglycemia some islets of male ZDF rats appear normal, while others are hypertrophic and display irregular outlines. A few islets show degenerative changes (Figure 3), characterized by β-cell vacuolation and hemorrhage into islet tissue from markedly dilated blood vessels. At this stage, insulin hypersecretion is apparent, although insulin mRNA levels remain normal (Tokuyama et al., 1995). After the onset of diabetes, as defined by severe, fasting hyperglycemia, islets are irregular with projections into the exocrine pancreas as a result of hyperplasia/hypertrophy and infiltration with inflammatory cells (Finegood et al., 2001; Janssen et al., 2001; Pickavance et al., 2003; Tokuyama et al., 1995). Following the onset of diabetes, insulin mRNA and insulin secretion are severely reduced (Tokuyama et al., 1995). Apoptotic bodies are also seen with higher incidence in islets and occur in conjunction with β-cell mass decline. As the β cell declines, fibroblasts are seen and the islets are eventually replaced with fibrous tissue with dilated and congested venules and capillaries (Figure 3).

By 12 weeks of age, when ZDF rats show fasting hyperglycemia, the β-cell mass was about half that present at 8 weeks (Finegood et al., 2001) and the rates of cell replication and neo-genesis of β cells were reduced while the rate of apoptosis was increased (Finegood et al., 2001; Tikellis et al., 2004). It is possible to prevent the reduction in β-cell mass by up-regulation of the β-cell growth factor, glucagon-like peptide 1, (Farilla et al., 2002; Sudre et al., 2002), and thereby, delay the onset of diabetes.

By 14 weeks, the degenerative changes in islets were more severe, and there was an almost complete absence of β cells within the islets that now consist predominantly of fibroblasts, collagen, and mononuclear cells (Figure 3). The irregular appearance of most islets is not only associated with rapid β-cell turnover with an overall net loss, but is most probably caused by the recruitment of mononuclear cells, substantially elevated numbers of fibroblasts, and collagen deposition.

The increased secretory demand placed upon the β cell during insulin resistance with sustained, raised glucose levels is frequently associated with a loss of insulin reserve without subsequent adequate repletion. This has been seen ultrastructurally as an increase in the ratio of immature to mature β-cell secretory granules, although an overall net decrease of these granules occurs (Chan et al., 1998). Secretory granule loss, demonstrated by immunohistochemical staining of proinsulin, is substantial in the human diabetic pancreas. This change is seen early in the pathogenesis of islet failure in disease models such as obese ZDF rats and quickly reverses on removal of the hypersecretory stimulus. Studies in which normoglycemia is restored have demonstrated reversal of β-cell exhaustion by the increased insulin content of the pancreas (Figure 4). This has been seen using agents such as insulin (Harmon et al., 2001), phlorizin (which promotes urinary excretion of glucose; Jonas et al., 1999; Kaiser et al., 2005b), and insulin sensitizers such as rosiglitazone (Finegood et al., 2001; Smith et al., 2000). In these studies, the stimulus for hypersecretion was inhibited and the cell allowed time to restore insulin reserves. It is notable that insulin mRNA is often not increased simultaneously with restored insulin reserve (Harmon et al., 2001), indicative, perhaps, that hypersecretion is a pathogenic factor in β-cell degranulation in prediabetes, or early diabetes, rather than a decreased ability to synthesize insulin.

Cultured islets from adult obese ZDF rats showed an increased basal insulin secretion and a reduced response to a glucose challenge compared with lean ZDF and (nondiabetic) Wistar rats (Sturis et al., 1994). The normal, pulsatile insulin release pattern was also perturbed. Furthermore, a hyperglycemia-dependent reduction in insulin mRNA was seen in ZDF islets (Harmon et al., 2001).

Changes in the expression of key glycolytic enzymes are implicated in contributing to β-cell dysfunction in ZDF rats. There is a reduction in the expression of mGDP and pyruvate carboxylase, both of which are involved in Goto-Kakizaki rat diabetes (MacDonald et al., 1996b; Tokuyama et al., 1995). ZDF rat islets are also desensitized to glucose-dependent insulinotrophic polypeptide (GIP; Lynn et al., 2001). Restoration of the leptin receptor in vitro elevates the expression of GLUT2, glucokinase, and preproinsulin toward normal levels (Wang et al., 1998). In restored cells, the addition of leptin to the culture improves glucose-stimulated insulin secretion and further increases glucokinase activity.

Many of the abnormalities in ZDF islets are thought to be lipid-induced. Adult ZDF rats are hypertriglyceridemic (Chirieac et al., 2004) and exhibit triglyceride accumulations in islets, which, interestingly, have been attributed to hyperglycemia (Harmon et al., 2001). The response of nondiabetic rodent islets to increased ambient free fatty acids is enhanced insulin production, higher glucose metabolism, and β-cell proliferation (Milburn et al., 1994). However, free fatty acids fail to elicit these changes in function in lean and obese ZDF islets (Hirose et al., 1995). This failure may be due to the activation of inducible nitric oxide synthase (iNOS; Shimabukuro et al., 1997), because inhibition of iNOS in vitro ameliorates the anomalous response to free fatty acids. Furthermore, treatment with thiazolidinediones has been shown to restore normal islet morphology (Buckingham et al., 1998) and reduce iNOS activity (Pickavance et al., 2003).

However, other pathways may be involved in islet fibrosis, because it has been shown that increased expression of angiotensin-converting enzyme 1 and 2 (ACE1 and ACE2) and the angiotensin 1 receptor has been observed in ZDF islets. When treated with ACE inhibitors (such as perindopril and irbestartan), ZDF islets show proinsulin repletion and amelioration of fibrosis and signs of oxidative stress (Tikellis et al., 2004). These effects are all independent of glycemic control, and since other antihypertensive agents have not improved islet morphology in ZDF rats, a direct action is indicated in this improvement.

Otsuka Long Evans Tokushima Fatty Rats

Otsuka Long Evans Tokushima fatty rats (OLETF) rats were inbred from a glucose-intolerant Long-Evans colony (Kawano et al., 1992). Diabetes developed relatively slowly in these animals, but by 20 weeks the rats were hyperinsulinemic and normoglycemic. Hyperglycemia developed by 40 weeks of age, with a greater incidence in males than in females (Rees and Alcolado, 2005). Characteristically, in adult rats there was mild to moderate obesity, mild hyperglycemia, hyperphagia, and renal disease, seen as glomerulosclerosis with mesangial matrix expansion and thickening of the glomerular basement membrane (Moran and Bi, 2006; Yagi et al., 1997). The cause of these changes is believed to be a deletion in the cholecys-tokinin-1 receptor gene, which leads to obesity and TTDM (Funakoshi et al., 1994; Tachibana et al., 1996). Dietary restriction or exercise prevented diabetogenesis in OLETF rats (Moran and Bi, 2006).

Kawano et al. (1992) classified islet histopathological changes in OLETF rats into three stages: (1) 6–20 weeks: Inflammatory cell infiltration and degeneration of islet architecture. (2) 20–40 weeks: β-Cell hyperplasia with proliferation of connective tissue. Late in this stage, β cells are insulin-exhausted through sustained insulin hypersecretion. (3) > 40 weeks: Islet atrophy is present caused by loss of β cells and fibrosis. Other islet endocrine cells remain unaffected.

Jia et al. (2005) provided further evidence of a chronic inflammatory process being intrinsic to islet failure. At 72 weeks, islets show marked immunohistochemical staining for the cytokines, tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and interleukin-6 (IL-6), which was not a feature during early diabetogenesis in these animals (Jia et al., 2005). Islet fibrosis was ameliorated by several molecules including acarbose (an α-glucosidase inhibitor), camostat (a protease inhibitor; Jia et al., 2005), bezafibrate (a hypolipidemic drug; Jia et al., 2004); and ramipril (an ACE inhibitor; Ko et al., 2004).

Psammomys obesus

Psammomys obesus (desert rat) is a gerbilid rodent found in North Africa and the Middle East and is used as a model of nutrionally induced diabetes (Kaiser et al., 2005b). Wild P. obesus consume a very low energy diet consisting mainly of low-calorie foliage, such as the leaves of the saltbush, Atriplex halimus. Their basal metabolic rate is approximately 60% that of other weight-and age-matched rodents (Degen, 1993). Normal rat chow is considered a high-energy food for P. obesus and causes hyper-glycemia in this species after 4–7 days. Colonies of diabetes-prone and resistant animals have been selected according to their high-energy diet response, allowing for the use of normoglycemic controls fed a high-energy diet. P. obesus is not hyperphagic by comparison with some rat strains—it modifies food intake to account for increased calorific value. Therefore, the natural history of diabetes in P. obesus shows similarities with several ethnic groups such as Pima Indians,Australian Aborigines, Asian Indians, and Mexican Americans in response to high-energy diet. TTDM is common in these native peoples since their domestic environments and therefore diets were Westernized (Lillioja and Bogardus, 1988).

The effects of a high-energy diet on P. obesus have been defined in four stages (A–D; Barnett et al., 1994; Lewandowski et al., 1998): (A) Animals are normoglycemic and normoinsulinemic; (B) animals are normoglycemic and hyperinsulinemic (indicating insulin resistance); (C) animals are hyperglycemic and hyperinsulinemic; and (D) animals are hyperglycemic and hypoinsulinemic. This inherent insensitivity to insulin and metabolic reliance on a low-calorie diet undoubtedly predisposes these gerbils to glucose intolerance. Hyperproinsulinemia marks the beginning of β-cell failure and associated hyperglycemia (stage C). As the amount of biologically active insulin decreases and glucose tolerance becomes increasingly compromised, β-cell failure is characterized by a loss of compensatory insulin hypersecretion (stage D).

Immediately following transfer to high-energy feeding (maintained for 22 days), loss of β cells occurs in P. obesus (Kaiser et al., 2005a), which quickly recovers, such that the β-cell mass exceeds that present before the change in diet. However, by stage D there is a marked increase in apoptosis and a reduction in β-cell mass (Bar-On et al., 1999). If the animals are returned to a low-energy diet, the reduction in β-cell mass recovers, indicating that the irreversible loss of β-cell mass is a late event (Kaiser et al., 2005b). It would appear that a reduction in β-cell insulin output and content is the primary lesion, rather than a reduction in cell numbers. Pancreatic insulin content is restored to 50% of that in normoglycemic controls when phlorizin is used to control hyperglycemia, despite the persistence of insulin hypersecretion (Kaiser et al., 2005b). These data suggest that it is insulin loss from β cells that is the major event in P. obesus diabetogenesis, in which islet morphology is characterized by β-cell degranulation and vacuolation. Furthermore, islets from P. obesus show greater susceptibility to a reduction in acute glucose-stimulated insulin secretion and intracellular insulin content after long-term exposure to supraphysiological glucose (Leibowitz et al., 2001). It is not fully understood whether this is due to a direct glucotoxic effect or a failure to cope with the metabolic demand initiated by high glucose levels.

Experimentally induced hyperinsulinemia, by use of a clamp, did not elicit change in the plasma glucose level of P. obesus, whereas diet-adjusted albino control rats became severely hypoglycemic under the same conditions (Ziv et al., 1996). This difference is not accounted for by compensatory hepatic glucogenesis. Indeed, while hepatic phosphoenolpyruvate carboxykinase activity decreased and glycogen was depleted as a response to hypoglycemia in the control rats, P. obesus showed no change in hepatic activity. Furthermore, albino control rats on a euglycemic, hyperinsulinemic clamp showed complete inhibition of hepatic glucose output whereas P. obesus showed only a slightly reduced output. These data indicate that the P. obesus gerbil is innately insulin insensitive.

Knockout and Transgenically Induced Diabetes

Genetically altered mouse models have provided mechanistic information on the function of specific proteins that may directly influence the development of TTDM or are relevant in the pathology of diabetic complications: Both mono- and poly-genetically modified mouse models have been used to these ends (Mauvis-Jarvis and Kahn, 2000, for a review).

Here, mice that express hIAPP are discussed and other transgenic mouse models are summarized in Table 2. More recently, a hIAPP-expressing rat (HIP) has been developed. HIP rats show pancreatic amyloidosis, insulin resistance, and reduced β-cell mass with resulting hyperglycemia (Butler et al., 2004).

Mice Expressing Human Amylin

Pancreatic amyloid is not a normal feature of rodent diabetes and transgenic mice that express hIAPP have been developed. Heterozygous expression of hIAPP in mice is not diabetogenic, whereas homozygous expression is associated with β-cell degeneration and hyperglycemia at 8 weeks of age (Janson et al., 1996). In the homozygous mouse, males show amyloid deposits at about 20 weeks of age. By contrast, females show a substantially different phenotype with diabetes occurring in about 20% of animals at 30 weeks of age.

Hyaline deposits are seen in islets and perivascular regions of homozygous hIAPP-expressing transgenic mice and are associated with both increases and decreases in insulin synthesis and secretion (D’Alessio et al., 1994; Janson et al., 1996).

Crossing heterozygous mice with an obese, insulin-resistant mouse (A^VY/a) has produced a strain with fasting hyperglycemia after 90 days, and severe hyperglycemia after 1 year (Soeller et al., 1998). This model is similar to the human condition because the expression of polygenic events links islet amyloid to obesity. Additionally, Hull et al. (2003) have shown that islet amyloid deposition in hIAPP-expressing mice is exacerbated by a high-fat diet, evidence that insulin resistance and insulin hypersecretion may be important for islet amyloid formation. This is consistent with the proposed pathogenesis of islet amyloid formation in humans (Zhao et al., 2003). Indications of a difference in response between mouse strains expressing hIAPP have been noted. For example, DBA2 mice expressing hIAPP show hyperglycemia and decreases in islet insulin content, whereas the BL6 mouse also expressing hIAPP does not develop diabetes even when maintained on a modestly high-fat diet (Hull et al., 2005b).

HIP Rat

HIP rats coexpress hIAPP with insulin and develop hyperglycemia and extrahepatic insulin resistance between 5 and 10 months, after initially showing reductions in first-phase insulin secretion (namely, postprandial, acute release of insulin) followed by a drop in plasma insulin to basal levels prior to the second phase of insulin secretion (prolonged hyperinsulinemia in response to postprandial hyperglycemia; Butler et al., 2004). These animals also showed hepatic insulin resistance and impaired fasting glucose.

At the onset of hyperglycemia, HIP rats show normal insulin levels but become hypoinsulinemic during the subsequent months of chronic hyperglycemia (Butler et al., 2004). Insulin deficiency in HIP rats was associated with both a reduction in intracellular β-cell insulin and a reduction in β-cell mass. Before the onset of hyperglycemia, β-cell mass was similar to that of normal rats, but was reduced by 60% at the onset of TTDM with an increase in β-cell turnover by comparison with normal rats. Furthermore, recently replicated HIP rat β cells show an increased tendency to apoptosis compared with older β cells and the overall frequency of β-cell apoptosis has been shown to be higher than that of replication or the formation of new β cells from ductal cells (Jones and Clark, 2001). Increased apoptosis precedes hyperglycemia and hypoinsulinemia, and worsens after onset.

Islets of HIP rats show progressive amyloidosis although this does not coincide with the appearance of apoptotic β cells. Therefore, if amyloidosis is related to β-cell apoptosis, it must be through an indirect mechanism. In contrast to other models, islet hypertrophy prior to hyperglycemia has not been reported in this model. Islets of hyperglycemic HIP rats exhibit severe insulin depletion, as assessed by immunohistochemistry (Matveyenko and Butler, 2006; Butler et al., 2004). These features of islet morphological changes in this strain are similar to those seen in human TTDM.

Insulin resistance is an important and interesting feature of this model because it was neither sought nor specifically induced; rather it is manifest as part of the development of the disease. Systemic insulin resistance in HIP rats has been confirmed by hyperinsulinemic, euglycemic clamp studies (Butler et al., 2004), which measure insulin sensitivity by determining the rate of glucose infusion when plasma insulin is maintained at a high level. Under these conditions, the rate of glucose infusion was shown to be reduced, which is indicative of insulin resistance. Furthermore, HIP rats also display an inability to suppress hepatic glucose production in a state of hyperglycemia (Matveyenko and Butler, 2006). The pathophysiological basis for insulin resistance in this model has not yet been identified.

Drug-Induced Diabetes

Streptozotocin-Induced Diabetes in Rats Fed a High-Fat Diet

A high-fat diet induces insulin resistance with compensatory hyperinsulinemia and when used in combination with a single dose of streptozotocin (STZ), which reduces β-cell capacity for compensatory insulin hypersecretion, results in hyperglycemia. The induced diabetic state continues indefinitely and is maintained long after exposure to STZ had stopped. In addition to a direct effect on glucose, which is characteristic of this model, other pathophysiological changes are seen, such as insulin resistance in adipose tissue (Reed et al., 2000) and diabetic kidney lesions such as glomerulosclerosis and proteinuria (Danda et al., 2005).

STZ is a β-cell toxin isolated from the bacterium Streptomycetes achromogens. It has a glucose moiety that is selectively taken up by glucose transporter-2 (GLUT2; Schnedl et al., 1994), which is expressed at high density in the β-cell membrane (Pang et al., 1994). STZ has been used successfully to induce models of both TODM and TTDM in genetically normal rats. STZ administered at 100 mg/kg of body weight results in a complete loss of β cells. A variation of the dosing regime involving administration of multiple, low doses (15 mg/kg) of STZ with Freund’s adjuvant induces a complement-involved loss of β cells (Ziegler et al., 1984), the pathogenesis of which is unclear. More recently, a lower dose of STZ (≤ 50 mg/kg) in conjunction with a diet rich in fat (40% fat by total dietary kcal) has been employed to induce a TTDM model in Sprague–Dawley rats (Danda et al., 2005; Reed et al., 2000; Sawant et al., 2004; Zhang et al., 2003; see Table 3). This model is a useful alternative to genetic models and may be more representative of Westernized diet-associated diabetes. Induction of diabetes in high-fat-diet-fed rats given STZ is characterized by a fasting hyperglycemia of about 20 mM. The degree of hyperglycemia is influenced by the composition of the high-fat diet and the STZ dose administered. However, the STZ-dosed, normal chow diet-fed controls remain unaffected, or develop a less marked hyperglycemia indicative of the synergistic action of high-fat diet (lipotoxicity) in STZ-induced diabetogenesis. Zhang et al. (2003) used a low STZ dose (15 mg/kg) by comparison with those used in other studies, but it is noteworthy that the 30% fat diet (by calories) was given for a substantial period (8 weeks) prior to dosing, further evidence of the complex interplay between the effects of different lipotoxic conditions and STZ. In their model, during the predose period, the rats developed insulin resistance to such a degree that a small reduction in insulin secretion was sufficient to induce hyperglycemia; STZ-dosed, normal chow-fed rats were unaffected. By contrast, Reed at al. (2000) administered a high-fat diet for a substantially shorter period (2 weeks) before dosing STZ at 50 mg/kg. STZ-dosed, normal chow-fed rats became slightly less hyperglycemic than STZ-dosed rats given the high-fat diet (see Table 3). Although hyperglycemia was slightly more pronounced in the Reed et al. model than in the Zhang et al. model (Table 3), a higher STZ dose was required to induce hyperglycemia in the former authors’ model. This finding implies that if a lower dose of, for example, 15 mg/kg is used after high-fat feeding for 2 weeks, hyperglycemia is unlikely to be induced. In the short-term, high-fat feeding may promote compensatory insulin secretion, whereas prolonged high-fat feeding is detrimental to β-cell function. Furthermore, Pascoe and Storlien (1990) developed a model of mild TTDM with slight hyperglycemia (8.6 mM) by dosing 2-day-old neonatal rats with 45 mg/kg STZ followed by feeding with a high-fat diet for 1 week when the animals had reached 8 weeks of age. With the exception of a defect in first-phase insulin secretion, no changes in plasma insulin or glucose levels were reported prior to the change in diet. Indeed, the interplay of possible synergies between high-fat feeding and STZ is important in consideration of the interplay between human obesity-related diabetogenic risks in the current worldwide obesity epidemic.

Islet hypertrophy is induced by high-fat-diet feeding (Hull et al., 2005a), indicative of altered β-cell function in response to plasma nutritional molecules. In contrast, STZ administration causes islet atrophy through β-cell loss, which depletes the central core of islet tissue leaving an ostensibly thickened layer of peripheral islet (non-β) cells (Zhang et al., 2003). However, the effect on β cells is mixed, those being lethally damaged becoming apoptotic, while others continue to function normally (Figure 5).

The dyslipidemia seen in this model is a feature of human TTDM and requires the combination of STZ and high-fat feeding because it is not induced when either STZ or a high-fat diet is used alone. The increased availability of glucose, with decreased utilization in insulin-resistant tissues, results in down-regulation of lipid oxidative pathways (Randle, 1998). Consequently, for rats given a high-fat diet, the blood lipid profile rises concomitantly with increasing glycemia and is likely to exacerbate insulin resistance by increasing lipid deposition in nonadipocytes, including β cells and skeletal muscle cells (Shimabukuro et al., 1998).

STZ-Induced β-Cell Dysfunction

STZ uptake into β cells results in the induction of a variety of intracellular mechanisms such as nitric oxide (NO) donation (Turk et al., 1993; Wada and Yagihashi, 2004), poly (ADP-ribose) polymerase (PARP) induction (Uchigata et al., 1982), DNA alkylation (Murata et al., 1999), and free radical generation (Wada and Yagihashi, 2004), which may be sublethal yet remain diabetogenic. STZ-induced β-cell toxicity is associated with a reduction in intracellular insulin, an effect that is distinct from β-cell cytotoxicity (Pieper et al., 1999). These mechanisms are discussed briefly below.

β-Cell toxicity induced by STZ is complex and may involve both genetic and nongenetic mechanisms. NO and other free radicals cause spontaneous chain reactions and result in non-specific molecular damage within the β cell following exposure to STZ. This is ameliorated by the NO scavenger, carboxy-PTIO (-2-phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide) but not by the NO synthase inhibitor, L-NAME (NG-nitro-L-arginine methyl ester; Wada and Yagihashi, 2004). STZ-induced DNA lesions activate PARP as a repair response, although overactivation can be a prodeath stimulus. Inhibition of PARP with nicotinamide prevents β-cell death (Wada and Yagihashi, 2004) and homozygous PARP-knockout mice are protected from STZ-induced diabetes, but not STZ-induced DNA damage (Peiper et al., 1999). Furthermore, in vitro studies have shown that methylnitrosourea, a nitrosamide derivative of STZ, induces DNA strand breaks and PARP induction to a similar degree as that seen when STZ is used, but without causing β-cell death (Wilson et al., 1988). This suggests that another step in STZ-induced β-cell death exists that is active beyond DNA damage and PARP induction.

STZ toxicity is associated with severely depleted intracellular insulin stores in treated β cells, due to rapid catabolism of nicotinamide adenine dinucleotide⁺ (NAD⁺), which is the substrate required in PARP activation (Cardinal et al., 1999). Subphysiological levels of NAD⁺ are associated with severe reduction in insulin production and other basal cellular processes. Indeed, while PARP overactivation is a proapoptotic stimulus, insufficient intracellular energy supply diverts apoptotic pathways to necrotic processes. Regardless of lethal effects, β cells that cannot synthesize insulin, due to a lack of ATP and NAD⁺, are extrinsically functionless. Hyperglycemia is the result of a substantial percentage of the total β-cell population being thus affected.

Surgically Induced Diabetes

Partially Pancreatectomized Rats

Partially pancreatectomized rats were among the first rodent models of diabetes (Lewis et al., 1950). Removal of approximately 90% of the pancreas results in delayed-onset hyper-glycemia between 12 weeks (Yomemura et al., 1984) and 10 months (Charreau et al., 1978), which may reflect the extent of pancreatectomy and/or strain differences (Kaufmann and Rodriguez, 1984). That a period of postoperative euglycemia exists before chronic hyperglycemia indicates that further loss of function of the remaining islet population occurs.

Distinct differences in β-cell gene expression are seen in 90% and 60% pancreatectomized rats by comparison with sham-operated (control) rats. Following a 90% pancreatectomy, rats initially show glucose intolerance and significantly increased proinsulin transcription per β cell (Orland et al., 1985). During this phase of glucose intolerance, the remaining β cells are hyperfunctional and able to compensate adequately to prevent fasting hyperglycemia. However, after 14 weeks they show fasting hyperglycemia and severe reductions in β-cell insulin mRNA. In contrast, after a 60% pancreatectomy, rats show an initial phase of glucose intolerance, and then recover to show normal, postprandial plasma insulin levels. These data suggest that a threshold exists for the degree of reduction in β-cell mass, which is compensated for by the remaining β cells, beyond which the remaining β-cell mass fails.

Changes in β-cell gene expression are related to the severity of hyperglycemia and are characterized by reductions in the expression of IAPP, islet-associated transcription factors, ion channels/pumps including GLUT2, glucokinase, mGPD, and pyruvate carboxylase (Laybutt et al., 2003). At 6 weeks after pancreatectomy, treatment with phlorizin to control hyperglycemia successfully ameliorates these gene expression changes, but at 14 weeks it does not (Laybutt et al., 2002). These findings indicate that in this model, irreversible diabetes develops with time and both the severity and the duration of hyperglycemia contribute to the changes observed in β-cell gene expression and, therefore, to reduced β-cell function.

The changes seen in this model are consistent with compensatory β-cell hyperfunction, that is, the β cells are hyperresponsive to arginine (an insulin secretagogue) in low-glucose conditions (Leahy et al., 1984; Rossetti et al., 1987). However, a relative reduction in insulin secretion is seen in response to glucose and nonglucose secretagogues at high glucose concentrations (Leahy et al., 1984; Rossetti et al., 1987). This reflects the high basal output of insulin per cell but an inability to increase insulin secretion to control acute rises in glucose levels. Indeed, this deficit in insulin appropriation is a feature of diabetes and prediabetic, glucose intolerance.

Histopathological lesions in partially pancreatectomized rats, such as islet hypertrophy (Laybutt et al., 2003) and fibrosis, correlate with severity of hyperglycemia (Clark et al., 1982). Insulitis, characterized by inflammatory infiltration into islets, has also been reported in association with the onset of hyperglycemia, and suppression of this inflammation has been shown to reduce the rate of islet degeneration (Lampeter et al., 1995).

Fetal Metabolic Programming

Permanent disturbances in metabolism and physiology in human adult life, such as obesity, TTDM, and cardiovascular disease, may be “metabolically programmed” by malnutrition during critical stages of intrauterine development (Barker, 1995). Typically, poor nourishment in utero results in a low birth weight, with bodily disproportions, but no further health implications until adulthood. This phenomenon was recognized as a significant factor for TTDM development in economically impoverished countries. Several methods have been developed for modeling this in rats. These include nutritional modulation in utero by reduced maternal dietary protein, ligation of the uterine artery (not discussed here), and in the neonate by dietary protein restriction or by feeding a high-carbohydrate milk diet (Barker, 1995). Organogenesis in the rat continues for a short period after birth, so that metabolic programming continues for a short period before weaning.

Neonates reared on high-calorie milk, fed by gastrostomy until weaning, almost immediately become hyperinsulinemic and develop chronic, severe insulin resistance with associated islet hypertrophy after 4 weeks (Vadlamudi et al., 1993). Obesity and glucose intolerance are also evident postweaning, although this regime alone is insufficient to induce fasting hyperglycemia. Several biochemical changes are seen in islets and accompany insulin hypersecretion. Hypersensitivity of these islets to glucose and other insulin secretagogues, such as glucagon-like peptide-1 (GLP-1), and increased basal insulin secretion have been confirmed in cultured islets. The exaggerated β-cell response to glucose is caused by increases in low hexokinase K_m activity, GLUT2 translation, and the activity of various glycolytic enzymes, namely mGPD and pyruvate dehydrogenase (PDH; Srinivasan et al., 2003).

When the availability of normal diet to pregnant Wistar rats is reduced, the offspring show reduced birth weights and are hyperphagic, hyperinsulinemic, and hyperleptinemic (Vickers et al., 2000, 2001). Postnatal, high-calorie feeding exacerbates these effects and also causes mild hyperglycemia (9.2 mM). This model is similar to human hyperphagia/obesity-induced TTDM and suggests that hyperphagia is an important part of metabolically programmed leptin resistance. Treatment with leptin during the neonatal period of metabolically programmed rats on a high-caloric diet has been shown to prevent the detrimental effects of metabolic programming (Vickers et al., 2005). This may be due, in part, to normalization of caloric intake due to leptin-induced satiation. Furthermore, the second-generation offspring of a malnourished dam are hyperglycemic when fed a high-fat diet after weaning (Martin et al., 2000), indicating an epigenetic generational transfer of diabetes susceptibility through metabolic programming.

Summary

Reliable and effective disease modeling is critical in the research and development of therapeutics for human disease. The models reviewed here demonstrate similarities, such as hyperglycemia, most often accompanied by dyslipidemia, but with numerous differences, such as the pathogenesis of islet degeneration and the plasma insulin profile. Since the presentation of TTDM in humans is clinically heterogeneous, we suggest that a particular rodent model may most closely represent the disease in some patients better than in others. Further disease delineation may identify markers and characteristics of patients that are represented by a particular model. For example, while the obese ZDF rat may represent Western “diabesity” where TTDM develops from obesity, the high-fat-diet-fed/STZ model may better represent Asian TTDM patients, whose BMIs are typically significantly lower than those of Western patients. Indeed, the plasma insulin profiles of these models indicate that insulin resistance is much more severe in the obese ZDF rat, and as this is readily assessable in patients, such as by hyperinsulinemic, euglycemic clamping, it should be possible to understand which models are more relevant to the clinical presentation of an individual patient’s TTDM and may be a step toward better glucose, and ultimately better disease, control.