A Review of Rodent Models of Type 2 Diabetic Skeletal Fragility

Abstract

Evidence indicating that adult type 2 diabetes (T2D) is associated with increased fracture risk continues to mount. Unlike osteoporosis, diabetic fractures are associated with obesity and normal to high bone mineral density, two factors that are typically associated with reduced fracture risk. Animal models will likely play a critical role in efforts to identify the underlying mechanisms of skeletal fragility in T2D and to develop preventative treatments. In this review we critically examine the ability of current rodent models of T2D to mimic the skeletal characteristics of human T2D. We report that although there are numerous rodent models of T2D, few have undergone thorough assessments of bone metabolism and strength. Further, we find that many of the available rodent models of T2D have limitations for studies of skeletal fragility in T2D, as the onset of diabetes is often prior to skeletal maturation and bone mass is low, in contrast to what is seen in adult humans. There is an urgent need to characterize the skeletal phenotype of existing models of T2D, and to develop new models that more closely mimic the skeletal effects seen in adult-onset T2D in humans.

Introduction

Type 2 diabetes mellitus (T2D) is chronic disease that affects multiple organ systems. Ample evidence indicates that T2D’s broad-ranging comorbidities do not spare the skeletal system. Older individuals with T2D have an increased risk of fracture, particularly at the lower extremities and hip (1–3), while several studies report a greater risk of vertebral fractures as well (2,4–7). In particular, meta-analyses reveal a 40% to 70% greater risk of hip fracture and a 20% increased risk of any clinical fracture in adults with T2D (1,3). These fractures are particularly problematic since patients with T2D exhibit compromised fracture healing (8) and poorer outcomes following hip fracture, including higher complication rates, longer hospital stays and worse functional recovery than non-diabetics (9–13). Altogether, these issues lead to greater healthcare costs among diabetics with a fracture (14).

The skeletal complications of T2D are of major concern given the global epidemic of obesity and T2D. Worldwide, cases of T2D have doubled from nearly 150 million in 1980 to almost 350 million today. In the US alone, nearly 30% of people over the age of 65 have diabetes, while 50% have pre-diabetes (15). Moreover, the incidence and impact of diabetes continues to grow. For instance, approximately 60% of the world’s diabetics live in Asia, the site of the fastest population growth rates in the world. Thus, the number of persons with diabetes is projected to increase two- to three-fold, with total diabetes-related health care expenditures expected to more than double in the next 25 years, reaching $490 billion worldwide (16). In the US, the largest increase in diabetes is predicted to occur in those over age 75 (17), the precise group that is already at highest risk for fractures.

In the United States, racial and ethnic minorities are disproportionately affected by T2D, with nearly a two-fold greater risk of diabetes in African-Americans and Hispanics compared to non-Hispanic Caucasians (18). This pattern is particularly troubling as Hispanics are the fastest growing sector of the U.S. population. The combination of population growth rates and these diabetes risk disparities among races/ethnicities are concerning, especially because diabetic skeletal fragility and increased fracture risk appear to affect races equally (2,4,19).

Currently, it is challenging to accurately assess fracture risk in patients with T2D. A key paradox is that increased risk of fracture is evident despite higher body weight and normal to high bone mineral density (BMD) in patients with T2D, factors that are generally associated with reduced fracture risk. Whereas low BMD is associated with increased fracture risk in T2D, on average diabetics suffer fractures at a higher BMD value than non-diabetics (20). Therefore, FRAX-based assessments of fracture risk underestimate fracture risk in those with diabetes (20,21). The underestimation of fracture risk by BMD could be in part explained by the increased rate of falls in diabetics, however, increased fracture risk persists after accounting for fall frequency (2). Thus, the reasons for increased skeletal fragility in T2D patients remain largely unexplained.

Given the incomplete understanding of what contributes to skeletal fragility in T2D, the value of animal models to study the pathogenesis and natural history of skeletal fragility in diabetes is without question. Ideal animal models would recapitulate the key metabolic and skeletal characteristics seen in humans with T2D. Additionally, animal models should be relatively inexpensive, easy to maintain, widely available, and amenable to genetic analyses and manipulations. Whereas there are numerous reviews on animal models of T2D and its complications (22–25), none have focused on the strengths and limitations of animal models of diabetes-related skeletal fragility.

Thus, in this manuscript, we aim to: 1) briefly review what is currently known of the effects of T2D on bone metabolism, mass, structure, and strength in human T2D; 2) describe and discuss current animal models for T2D and skeletal fragility; and 3) critically compare/contrast the characteristics of animal models to the clinical presentation of bone outcomes in patients with T2D. This review will provide a new perspective on the strengths and limitations of animal models of T2D, as well as shed light on current challenges in the study of diabetic skeletal fragility, and future areas for research.

Bone metabolism and structure in adults with T2D

Human T2D is characterized by inadequate beta-cell response to insulin resistance, leading to hyperglycemia. Increased body weight, low physical activity levels, and advanced age are all associated with T2D, although there is a growing incidence of T2D among overweight children and young adults as well (26–28).

Compared to non-diabetics, adults with T2D have normal or increased areal bone mineral density (aBMD) at the spine and hip (29). Bone formation is consistently lower in patients with T2D compared to non-diabetics, as evidenced by lower serum osteocalcin and procollagen type 1 N-terminal propeptide (P1NP) levels (30–37), as well as reduced histologic measures of mineralizing surface and bone formation rate (34). Reports of the effects of T2D on bone resorption are less consistent, with studies reporting either no difference or reduced bone resorption markers in patients with T2D (30,34,36–39). Serum PTH tends to be lower (37,40) while levels of the Wnt-signaling antagonist, sclerostin, are higher in patients with T2D (38,39,41–43).

Though generally limited to small sample sizes, cross-sectional studies have been used to examine cortical and trabecular bone density at the hip and spine (44), as well as peripheral skeletal sites (distal radius and tibia) in adults with T2D (35,44–48). The most consistent observation is that trabecular bone density in T2D is generally similar to or significantly greater than that of non-diabetic controls (35,44–48). The effects of T2D on cortical bone are more variable, with some studies reporting no differences (35,44), whereas others find deficits in bone size and/or cortical bone structure in T2D (45,46,48). Burghardt et al (46) first reported increased cortical porosity in postmenopausal women with T2D. However, a recent study found that cortical porosity is increased only in postmenopausal diabetics with a prior fragility fracture, and that diabetics without a prior fracture had similar cortical bone density, porosity and microarchitecture as non-diabetic controls (48). Additional studies are needed to characterize bone microstructure in T2D, particularly in men and in non-Caucasian populations.

The accumulation of advanced glycation end-products (AGEs) in the organic bone matrix, and associated deleterious effects on bone mechanical properties, has been proposed as a potential mechanism underlying skeletal fragility in T2D (49). Urinary and serum pentosidine do not differ between diabetics and non-diabetics (50,51), but higher urinary pentosidine levels are associated with incident vertebral fractures among those with T2D (51). A recent study performed in vivo reference point indentation of the mid-tibia diaphysis and showed that postmenopausal women with T2D had worse indentation properties (i.e., greater indentation depth) than non-diabetic controls (52). It is notable, however, that data on AGEs in diabetic human skeletal tissue are scarce. More work is necessary to determine the contribution of AGEs to skeletal fragility in T2D.

Animal Models of T2D

The rodent models of T2D discussed below are organized in a four-tiered system that reflects the etiology and timing, as well as the physical phenotype associated with diabetic symptoms in rodent models. The mouse and rat models are classified as (i) spontaneous or diet-induced, (ii) single gene or polygenic etiology, (iii) obese or lean body type, and (iv) by the timing of T2D onset, either before or after skeletal maturity. This classification system was chosen for a number of reasons. Although diet-induced diabetic models have not received much attention in studies of diabetic skeletal fragility, in theory these types of models best simulate human T2D etiology in most regions of the world and differ from spontaneous or genetically-induced models of T2D (24,25,53). Spontaneous models are then subdivided into single gene or polygenic etiologies, as it is expected that the polygenic models are more likely to reflect the complex genetic contributions to T2D in humans while the mouse models with single gene mutations provide a robust platform to test specific mechanistically-oriented hypotheses. Next, we identified models as lean or obese. Obese models of T2D are commonly used in studies of diabetes-related skeletal fragility, but the growth of T2D in low BMI patients in Asia and other regions reinforces the importance of this distinction (54). Finally, during growth the skeleton predominantly experiences modeling processes while the adult skeleton predominantly experiences remodeling processes. In general, skeletal maturity is considered the transition between these two phases. We used a classification relative to the age of skeletal maturity because we hypothesize that the overall impact of T2D on bone mass and microarchitecture will strongly depend on the age of onset. Hyperglycemia or frank diabetes before or after skeletal maturity may disrupt normal bone metabolism in both cases but the outcomes may be very different because disruption in the former affects bone mass acquisition and development, whereas the latter affects skeletal maintenance activities.

Skeletal maturity ranges were defined as 10–12 weeks in the mouse (55) and 15–17 weeks in the rat (56) based in part on the slowing of skeletal growth. In mice, the timing of “peak bone mass” is difficult to define, as age-related trabecular bone loss begins as early as 8 weeks of age (55), while cortical bone mass increases until approximately 4 months of age (57,58). In this review, rodent models that developed diabetes before skeletal maturity or coincident with skeletal maturity were grouped together as “pre-skeletal maturity” because bone acquisition most likely occurs in either a pre-diabetic or diabetic condition.

Normal and diabetic fasting glucose levels differ in rodents and humans (Table 1). Unfortunately, criteria for defining diabetes in animal models are rarely provided in primary reports or review articles (23–25,53,59–62). Fasting glucose levels between 100 and 199 are common among mouse strains, even after treatment with a high-fat diet. This range is not typically associated with diabetic symptoms such as polyuria and polydipsia (63). However, mouse strains with fasting glucose levels >250 mg/dl (diet-induced or spontaneous) do present with these and other symptoms (25,63–65). Therefore, we defined diabetes in rodents as a fasting glucose level >250 mg/dl or greater, whereas normal fasting glucose levels and pre-diabetes were defined as <200 mg/dl and between 200 and 250, respectively. In the text below, unless otherwise noted, we report or refer to fasting glucose levels.

Table 1.

Fasting plasma glucose levels (mg/dl) for normal, pre-diabetes and diabetes: comparison of human and rodent

	Human	Rodent (23,63)
Normal	<100	<199
Prediabetes	100–124	200–249
Diabetes	> 125	> 250

A summary of the key metabolic and skeletal characteristics of murine and rodent models of T2D are found in Tables 2 and 3, respectively, while a more detailed review of each model is presented below. The summaries presented here are based on reviews of the literature using PubMed, Google Scholar, and Medline and only include published data.

Table 2.

Mouse models of T2D

Diet-Induced
Polygenic Models
Body composition	Age of onseta	Animals	Parental Strain	BMD status	Biomechanical Properties	Bone turnover markers	Histo morphometry	Comments	Key references
Obese	Depends on start of high-fat diet	C57Bl/6J	N/A	↓ 1,3,4	—Max Load1 —Stiffness1 ↓Yield Load3 ↓Eapp1	—OCN	↓BFR/BS ↑ES/BS	Susceptible to diet- induced obesity, but resistant to diabetes. Effect of high fat diet depends on age.	(63,67,72,76, 78)
Spontaneous
Single Gene Mutation Models
Lean	Pre- maturity	Muscle IGF- 1R–lysine- arginine (MKR)	FVB	—1	↓Max Load1 ↓Stiffness1	↑TRAP	↑Er.Pm ↓MARperiosteal	Dominant negative mutation in muscle- specific IGF1R	(82)
Obese	Pre- maturity	ob/ob	C57Bl/6J	↓1	↓Max Load1	↑uDPD	↓BFR/BS —Ob.N/BS ↑Oc.N/BS	Leptin deficient	(93,97)
		db/db	C57Bl/6J	↓1,2	↓Max Load1 ↓Stiffness1	—TRAP ↓OCN ↓CTX	↓BFR/BS —Ob.N/BS ↑Oc.N/BS	Lepr deficient	(93,94,97,107)
		Yellow Kuo Kondo (KK/Ay)	KK	↓1					(110,129)
Polygenic Models
Obese	Pre- maturity	Tallyho	SWRb	↓4	↑Max Load1	↓OCN			(85,113,114)
		M16	ICR	↑4					(118)
		Nagoya- Shibata- Yasuda (NSY)	JcI:ICR					Male T2D bias, moderate obesity	(120)
		Tsumura Suzuki Obese Diabetes (TSOD)	ddY					Male T2D bias	(124,125)
		Kuo Kondo (KK)	ddY						(110,129)
	Post- maturity	New Zealand Obese (NZO)	New Zealand Black (NZB)					Autoimmune complications	(25)

Symbols: ↑ significant increase compared to controls, ↓ significant decrease compared to controls, -no change compared to controls,

¹femur,

²tibia,

³vertebra,

⁴total body

^aAge of onset is determined relative to skeletal maturity in mouse and rat, as is explained in the text.

^bTallyHo has 86% genetic similarity to the SWR mouse.

Abbreviations: Apparent elastic modulus (E_app), osteocalcin (OCN), Tartrate-resistant alkaline phosphatase (TRAP), urinary deoxypyridinoline (uDPD)

Table 3.

Rat models of T2D

Diet-Induced
Polygenic Models
Body composition	Age of onseta	Animals	Parental Strain	BMD status	Biomechanical Properties	Bone turnover markers	Histo morphometry	Comments	Key references
Obese	Depends on start of high-fat diet	Israeli Sand Rat (Psammomys obesus)	N/A				—Ob.S/BS ↑Oc.N/BS	Normal lab diet induces obesity and T2D	(144,145,190 ,191)
Spontaneous
Single Gene Mutation Models
Obese	Pre- maturity	Zucker Diabetic Fatty (ZDF)	Zucker	↓1,3	↓Max Load13 ↓Stiffness13 —Eapp1,3 —Toughnessapp1	↓OCN, ↓P1NP ↓CTX —uDPD	↓ or—BFR/BS	Lepr mutation, Male T2D bias	(61,146,151)
Polygenic Models
Lean	Pre- maturity	Goto-Kakizaki (G-K)	Wistar	↓1,2,3	↓Max Load12 —Eapp3	↓OCN ↑TRAP	↓BFR/BS		(164,165)
	Post- maturity	Diabetic Torrib (SDT)	Sprague-Dawley	—1	↓Max Load1 ↓Stiffness1	—OCN	↓BFR/BS ↓Ob.S/BS ↓Oc.N/BS	Male T2D bias	(171,172)
Obese	Pre- maturityc	Zucker Diabetic (ZDSD)	DIO Sprague Dawley and ZDF	↓1,2	↓ Yield Load13 ↓Max Load12 ↓Stiffness12 ↓Eapp1 ↓Eapp2 ↓Ult stress2			Females may develop adult T2D, male T2D onset near or at maturity	(61,173,174)
	Post- maturity	Otsuka Long Evans Tokayashima Fatty (OLETF)	Long-Evans	-	↓ Yield Load	↓OCN —TRAP		Develops impaired glucose tolerance around 24 wks, Male bias	(59,179,189)

Symbols: ↑ significant increase compared to controls, ↓ significant decrease compared to controls, —no change compared to controls,

¹femur,

²tibia,

³vertebra,

⁴total body

^aAge of onset is determined relative to skeletal maturity in mouse and rat, as is explained in the text.

^bDiabetic Torri rat BMD and turnover markers are similar to controls at 20 wks, but formation decreases and resorption increases at 36 wks

^cZDSD male rat spontaneously develops T2D just prior to or at maturity, limited data suggest females develop T2D at or after skeletal maturity

Abbreviations: Apparent elastic modulus (E_app), osteocalcin (OCN), Tartrate-resistant alkaline phosphatase (TRAP), urinary deoxypyridinoline (uDPD)

I. Murine Models

A. Diet-Induced Models of T2D

Polygenic/Obese

C57Bl/6J Mouse

The C57Bl/6J mouse is perhaps the best-studied model of diet-induced obesity, but it is not a good model of diabetes because this model never develops frank diabetes. The C57Bl/6 is reviewed here to be thorough and provide information on the best-studied mouse model of high-fat diet induced obesity.

Although not all studies agree (66–68), a high-fat diet appears to consistently induce impaired glucose tolerance with glucose levels reaching approximately 200 mg/dl (63,69–72). Impaired glucose tolerance in C57Bl/6J appears to result from poor insulin secretion (71) but insulin resistance also develops with increasing age and concomitant exposure to a high-fat diet. Impaired glucose tolerance is more severe in males than females (71,73). Renal morphological and functional changes have been reported after exposure to a high fat diet (74). Moderate peripheral nerve sensation deficits have also been noted in females exposed to high fat diet but retinopathy-like changes have not (75).

Although the C57Bl/6J mouse never develops diabetes, a high-fat diet and obesity impact the skeleton. The high-fat diet does not affect bone length in C57Bl/6J mice (72,76) but it appears to have adverse effects on trabecular bone compartments (67,76–79), though recent evidence suggests that the effect of the high-fat diet is more pronounced in skeletally immature compared to skeletally mature mice (67). Lower trabecular bone volumes are primarily the result of increased osteoclast resorption, as indicated by serum biomarkers and histomorphometry (76–79), though bone formation may also be decreased (79). Interestingly, recent data indicate that while body weight and fasting glucose levels can return to normal after a return to a normal diet, trabecular bone loss prior to skeletal maturity may not be recoverable without some other type of intervention (67). The potential implications of those results are obvious considering the childhood obesity problem in the United States and other countries.

The effects of a high-fat diet on cortical bone in C57Bl/6J are less clear. Long-bone diaphyseal cortical bone thickness and area have been reported to increase (68), decrease (67,72,80), and not change (76,77) after long-term exposure to a high-fat diet. Limited data on the spine suggests that the cortical shell thickness is unchanged in obese C57Bl/6J mice compared to lean controls (67,78). In spite of these disparate results, one consistent trend is that diet-induced obesity has a greater effect on trabecular bone volume than cortical bone mass and structure. Generally, femoral whole bone properties in obese mice are similar to controls, whereas material properties such as the estimated elastic modulus, bending strength, and fracture toughness are significantly lower compared to lean controls (68,72). Reduced material properties in the femur may result from increased accumulation of advanced glycation end-products in the obese C57Bl/6J mice relative to controls (72). In the lumbar spine, vertebral body stiffness, yield force, and energy-to-maximum force are significantly decreased in C57Bl/6J mice exposed to a high fat diet at a young age compared to lean controls (67).

B. Spontaneous Models of T2D

Single Gene/Lean/Pre-Skeletal Maturity

Muscle IGF-1R–lysine-arginine (MKR) Mouse

The MKR mouse was developed by introducing a point mutation of lysine to arginine, producing a dominant negative mutant of the insulin-like growth factor 1 receptor (IGF-1R), specifically in the skeletal muscle (81). MKR mice are lean, but have naturally occurring hyperinsulinemia and insulin resistance in the muscle from birth (82). Hyperinsulinemia is prominent by 2 wks, and is followed by an increase in insulin secretion that leads to insulin resistance by 3 wks (83). By 7–8 wks, MKR show obvious hyperglycemia and hyperlipemia, with blood glucose levels reaching ~360 mg/dl (82,84,85). MKR mice lose 20% body weight during the first 4 wks of age, and continue to lose another 10% throughout life compared to wild type controls. Thus, these mice develop T2D early and show severe muscle insulin resistance.

One study has reported skeletal characteristics in MKR (82). MKR mice femurs have decreased stiffness, decreased failure load, but increased post-yield displacement compared to controls at 16 wks age. They also have increased femoral slenderness, but no difference in length compared to controls. Tissue mineral density in MKR does not differ from controls at any point between 3–16 wks. At 16 wks age, BV/TV in the distal femoral metaphyses of MKR mice is significantly lower than controls. There is decreased osteoblast activity (e.g. percent labeled perimeter) on the periosteal surface from 3–8 wks, and increased osteoclast activity (e.g. percent eroded perimeter) at both endosteal and periosteal surfaces of MKR femurs from 8–16 wks.

Single Gene/Obese/Pre-Skeletal Maturity

Ob/ob Mouse

The ob/ob mouse is a model of severe obesity resulting from a spontaneous inactivating mutation in the leptin gene. This autosomal recessive mutation was originally detected in a non-inbred mouse stock at the Jackson Laboratories, and was then backcrossed into the C57BL/6J strain (86). Weight of ob/ob mice increases rapidly starting at 2 wks of age, and can reach up to 3-fold the weight of wild-type controls (53,87). Hyperphagia, insulin resistance, and hyperinsulinemia are all evident at 3–4 wks of age, with obesity evident by 4 wks (87). Even after obvious hyperglycemia detected at 4 wks, blood glucose levels continue to increase until reaching a peak at 12–20 weeks of age with glucose levels reaching up to 400 mg/dl (87). After this peak, blood glucose levels fall until reaching a normal level at older age. Other phenotypic characteristics of ob/ob include hyperlipidemia, low physical activity, low body temperature and impaired thermogenesis, impaired wound healing, infertility, deficits in motor and sensory nerve conduction velocity, and retinopathy (87–91).

Ob/ob mice are used as a general model for insulin resistance and obesity, and the ob/ob mutation has been incorporated into the BTBR and C57BLKS/J backgrounds, among others. Although ob/ob mice may serve as a useful model for the study of pre-diabetes, insulin release capacity is high through its life, hyperglycemia decreases after 24 weeks, and other diabetic complications (e.g. reduced kidney function) are not as prominent as in other models (e.g. db/db mice). There is also a lack of complete β-cell failure in the model, indicating that diabetes is not very severe (92). Moreover, because leptin itself directly influences bone metabolism, the skeletal phenotype of the ob/ob mouse cannot be attributed solely to the observed metabolic changes. Hence, it seems that the ob/ob mouse is not the best model to represent later stages of human T2D.

In contrast to initial reports of high bone mass in the ob/ob mouse (93), subsequent studies show a complex skeletal phenotype with ob/ob having higher trabecular bone mass in the lumbar vertebrae, but lower trabecular and cortical bone mass in the long bones (94–96) as well as shorter femurs compared to controls (94,96,97). Furthermore, ob/ob mice have decreased biomechanical properties (e.g. maximum load) as determined by 3-pt bending of the femur (94–96). They also have lower bone mass and markedly reduced bone formation in vertebrae and long bones compared to wild-type controls, and this deficiency can be corrected with leptin treatment (93,97).

Db/db Mouse

The db/db mouse is a model of severe type 2 diabetes resulting from an autosomal recessive mutation of the db gene, which codes for the leptin receptor, in the C57BL/KsJ mouse strain (98). Db/db mice are spontaneously hyperphagic and secrete excessive insulin, making them obese, insulin resistant, hyperinsulinemic, and hyperglycemic within 4 wks age (98). Plasma insulin starts to increase within 2 wks of age and blood sugar increases within 4–8 wks. Hyperglycemia peaks at 12–16 wks, reaching up to ~400 mg/dl while fasting (99). There is a severe depletion of insulin-producing β -cells, and ultimately, db/db mice do not live longer than 32–40 wks (98).

Diabetic symptoms are much more severe in db/db mice than in ob/ob mice. Clinical symptoms are more severe in males than in females. db/db mice have decreased body length (~5% shorter than controls) (100), develop abnormal kidney morphology, have deteriorated cardiac function (including ischemia, decreased blood pressure, and abnormal blood circulation), glomerular hypertrophy, and deteriorated retinal neuronal function (101–103).

The skeletal phenotype of the db/db mouse is difficult to characterize. Although there is consensus that long bones in db/db mice are 2–7% shorter than wild-type controls (104–107), there are conflicting data regarding other aspects of bone structure. Several studies report worsened bone properties compared to controls. For example, one study reported that 5 wk old db/db mice have decreased femoral cortical area and thickness compared to controls (104). Another study conducted on 11 wk old mice reports decreased trabecular and cortical bone volume, trabecular number, and trabecular thickness in the tibia, normal trabecular and cortical bone volume in the vertebrae, and decreased vertebral trabecular and cortical thickness (107). In contrast, another study reported increased trabecular number and bone volume fraction in femurs and vertebrae as well as increased bone formation rate in 3- and 6-mos old db/db mice (93). Mechanical testing results indicate that db/db mice have lower femoral failure load and stiffness than controls (107). Additionally, serum osteocalcin and CTX levels are lower in db/db (97,107), and there is reduced bone formation rate in the lumbar vertebrae femur, as recently reported by Turner et al (97). As with the ob/ob mouse, since leptin has direct effects on the skeleton, one limitation of the db/db model is that the observed skeletal phenotype cannot be attributed solely to its metabolic disturbances.

Yellow Kuo Kondo (KK/Ay) Mouse

The Yellow KK (KK/Ay) mouse was produced from the KK mouse (reviewed below) by transferring the Agouti gene into the KK mouse in order to develop a line with frank diabetes (108). The KK/Ay mouse develops severe obesity, insulin resistance, hyperglycemia, and hyperinsulinemia by 8 wks of age. By this time, plasma glucose reaches about 300 mg/dl, and increases up to 400–450 mg/dl by 16 wks (109). Hyperglycemia and hyperinsulinemia progress with age and no diet supplementation is needed for the mice to continue in a diabetic state (25). The KK/Ay mouse is often utilized as a model of diabetic nephropathy. Renal characteristics show thickening of the glomerular capillary walls, mesangial cell proliferation, and advanced glycation end-products accumulating in the glomerular mesangial regions (108).

Very little information is available on the skeleton of KK/Ay mice, with a single study reporting that proximal femur BMD is less in KK/Ay than controls at 18 wks of age (110).

Polygenic/Obese/Pre-Skeletal Maturity

TallyHo Mouse

The Tallyho mouse is a model of early onset, naturally occurring type 2 diabetes mellitus and obesity. This inbred polygenic model, developed at the Jackson Laboratory, was derived from selective breeding of outbred Theiler Original mice that were spontaneously hyperinsulinemic and hyperglycemic (111,112). The genetic background is mixed, but is 86% homologous with SWR mice. TallyHo mice are fatter and weigh more than controls by 4 wks of age (85,113,114). Body weight continues to increase with age, but obesity in these mice is less severe than some other diabetic models (85). Increased plasma triglycerides, cholesterol, and fatty acid levels are noticeable from 4–8 wks of age (85). Diabetes is present only in male mice and plasma glucose levels approach 300–350 mg/dl at approximately 10+ wks of age (24). Typical complications of diabetes, such as retinopathy and neuropathy do not occur in TallyHo mice. Some renal changes have been noted in TallyHo mice but they occur prior to the onset of diabetes and are similarly seen in SWR mice (115).

Other characteristics of the TallyHo mouse include enlarged pancreatic islets, hyperleptinemia, and poor wound healing (85,114,116). Additional phenotypic characteristics specific to male TallyHo mice include atypical pancreatic islet architecture and endothelial dysfunction (85,112,113).

Investigations of the skeletal phenotype in TallyHo mice have utilized two different control strains. Won et al (113) used C57Bl/6J mice to show that male TallyHo, but not female, have reduced BMD at 8 and 12 wks of age. TallyHo also exhibited reduced serum osteocalcin, and lower osteoblastogenic and higher osteoclastogenic gene expression patterns in bone marrow cells compared to C57Bl/6J (117). Devlin et al (117), using SWR mice as controls, showed that despite their greater body weight, male Tallyho mice have reduced total body BMD as well as lower trabecular bone volume at the distal femur. Femoral 3-pt bending showed that male Tallyho mice have a higher maximum load but less post-yield deformation than SWR controls (117).

M16 Mouse

The M16 mouse is a polygenic model of early onset obesity and moderate hyperglycemia that resulted from long-term selection of ICR outbred mice for accelerated weight gain in the first 3 to 6 wks of life (53,118). M16 mice are larger than ICR controls at all ages. They have increased fat cell size and number, body fat percentage, and organ weights (53,119). M16 mice are hypercholesterolemic, hyperinsulinemic, and hyperleptinemic by 6 wks of age (118). With increasing age insulin secretion begins to decrease and glucose levels begin to increase most notably in males, and by 8 wks age male M16 mice have an average glucose level just below ~250 mg/dL (pre-diabetic range), while females are at ~150/dL (normal range) (118). After 8 wks of age, these glucose levels begin to decrease (118). Common human diabetic complications such as nephropathy, neuropathy, and retinopathy have not been reported in M16 mice.

A potentially unique trait of the M16 mouse is that males and females have greater total body BMD than controls at 8 wks of age (118). Because the majority of hyperglycemic and diabetic rodent models exhibit low BMD, it is unclear if high BMD in the M16 mice is due to the development of mild-moderate hyperglycemia instead of frank diabetes or to other factors, such as their greater body weight. No other information exists on the skeletal phenotype of M16 mice.

Nagoya-Shibata-Yasuda (NSY) Mouse

The NSY mouse is an inbred strain developed by selective breeding for increased glucose levels and hyperinsulinemia from outbred JcI:ICR mice (120). These mice exhibit mild obesity, with visceral and abdominal fat accumulation (25). Fasting glucose levels remain in the normal range as the mice age, but glucose challenges reveal impaired glucose tolerance (120). Impaired glucose tolerance after a glucose challenge worsens with age and is more prevalent in males (98%) than in females (31%) (120). NSY mice have increased insulin content in the pancreas, but no change in pancreatic islet size (120). The NSY model is useful to investigate age-dependent characteristics of impaired glucose tolerance. Retinopathic and neuropathic complications have not been reported for the NSY mouse. Some diabetic nephropathy-like changes were reported after the initial introduction of the NSY mouse (121,122), but it is unclear if these changes are strain-dependent, disease-dependent, or both (123). There is no information on the skeletal phenotype of NSY mice.

Tsumura Suzuki Obese Diabetes (TSOD) Mouse

The TSOD mouse strain was developed in 1992 through selective inbreeding of obese male ddY mice, which is an inbred strain from Germany. This mouse is a polygenic model of obesity and impaired glucose tolerance. Male TSOD mice develop insulin resistance, polyuria, and polydipsia by 8 weeks of age (124–126). Obesity continues to develop until 12 months of age (53). Pancreatic islets are hypertrophic and β-cell mass increases (124). Compared to the Tsumura Suzuki Non Obese (TSNO) control, the TSOD mouse is hyperinsulinemic (124–126). However, the TSOD mouse exhibits a complex phenotype in which the fasting glucose levels remain between 100 and 200 mg/dl as the animal ages (127), but glucose challenges push plasma glucose levels above 300 mg/dl 120 min after the introduction of the glucose. In addition, diabetic renal changes and peripheral neuropathy have been noted in the TSOD mouse (126). Diabetic retinopathy has not been noted in this mouse. There is no information on the skeletal phenotype of TSOD mice.

Kuo Kondo (KK) Mouse

The Kuo Kondo (KK) mouse is a polygenic model of moderate obesity and T2D that originated in Japan (25). The exact time of onset of diabetes is unclear (128), but by 8 weeks of age the KK mouse has achieved moderate obesity, insulin resistance, and hyperinsulinemia (25), with blood glucose reaching 250 mg/dl by 16 wks (129). Insulin resistance is similar to that in db/db mice, although KK mice are less obese (129) at 9–12 months of age (130). Male KK mice have more evident obesity and hyperinsulinemia than females (108). The KK mouse is also prone to developing albuminuria and progressive kidney disease, making this strain useful for the study of nephropathy (109). Some retinal changes occur with diabetic onset although this condition is not well studied in the KK mouse (131). Diabetic neuropathy has not been reported. To our knowledge, there have been no studies conducted on skeletal characteristics of the KK mouse.

Polygenic/Obese/Post-Skeletal Maturity

New Zealand Obese (NZO) Mouse

The inbred NZO mouse strain is a polygenic model of obesity and T2D, developed by selective inbreeding with parents selected for their agouti coat color (132,133). These mice gain weight within the first 10 wks due to hyperphagia and reach peak weight at 48 wks of age (134). These mice show progressive impaired glucose tolerance, and approximately half of the male mice develop frank diabetes, with blood glucose levels reaching 300–400 mg/dl by 20–24 wks of age (135). NZO mice have a tendency to develop autoimmune diseases. Specifically, they develop glomerular proliferation and glomerulosclerosis (136) that are partly consistent with lupus (136–138). Neuropathy and retinopathy do not appear to develop. These mice are not commonly selected as a murine model, but are useful to investigate relationships between obesity, hyperglycemia, and autoimmune diseases (25). To our knowledge, nothing is known about the skeletal phenotype of the NZO mouse.

II. Rat Models

A. Diet-Induced Models of T2D

Polygenic Mutation/Obese

Israeli Sand Rat (ISR)

The Israeli Sand Rat, which is actually a gerbil (Psammomys obesus), is classified here as an early onset polygenic obesity model, but the model is available for use after skeletal maturity as well. The ISR is very sensitive to changes in diet due to its adaptation to the desert environment. A shift to a lab-based high-energy diet induces obesity, hyperinsulinemia, and subsequent diabetes. With 3–4 wks of a high-energy diet, blood glucose levels can reach up to 360–540 mg/dL (139). The dietary change can be implemented at any age (139). Later stages of disease progression are similar to humans with beta cell degradation, severe insulin deficiency, nephropathy, peripheral sensation loss, body weight loss, and consequent death (65,140,141). Unlike humans, the ISR does not experience retinopathy (142). The unique sensitivity to a dietary shift suggests that Psammomys is an example of the Thrifty Gene Hypothesis (141,143).

Studies performed in the late 1980s and early 1990s suggest that vertebral trabecular bone area fraction and hind limb cortical bone thickness in four-month old diabetic and non-diabetic ISR and controls are similar (144,145). These earlier studies were limited in sample size and the diabetic condition was limited in duration (less than 3 months) so it is unclear whether the ISR experiences diabetes-associated skeletal changes (144).

B. Spontaneous Models of T2D

Single Gene/Obese/Pre-Skeletal Maturity

Zucker Diabetic Fatty Rat (ZDF)

The Zucker Diabetic Fatty Rat model, like the db/db mouse, has a leptin receptor deficiency caused by an amino acid substitution (146). Diabetic symptoms become overt around 9 wks of age and only occur in males, with fasting plasma glucose levels approaching 300 mg/dl. The progression of diabetes includes early hyperinsulinemia (due to insulin resistance), later hyperglycemia, followed by decreased insulin secretion along with moderate obesity (at all stages). Consistent with the severe diabetic condition, the ZDF rat develops peripheral sensory deficits (147), renal hypertrophy, glomerulosclerosis, and proteinuria but lacks other classic changes such as renal mesangial matrix expansion (64,148). Diabetic retinopathy has not been reported for the ZDF rat.

The skeletal characteristics of the ZDF rat have been well studied. Compared to controls, femoral length is shorter and femoral diameter smaller in ZDF rats (149). BMD, microCT-based structural measurements, and mechanical properties are similar to controls at 7 wks (150), just prior to the onset of overt diabetic symptoms. Thereafter, BMD (61,151) and structural measurements begin to decline relative to the controls in the hind limbs and spine.

A few studies have shown that the bone formation markers P1NP and OCN are lower in ZDF, partly contributing to delayed fracture healing in these animals, but other studies have reported normal levels of these markers (149,152,153). Similarly, resorption biomarker results are not in agreement. Urinary DPD measurements suggest normal resorption in ZDF rats compared to controls, but measurements of CTX indicate increased resorption (149,151,152). Limited histomorphometry data indicate that bone formation is lower at the lumbar vertebrae (but normal in the proximal tibia) in ZDF rats compared to non-diabetic controls, while osteoclast number tends to be higher in ZDF (151).

Polygenic/Lean/Pre-Skeletal Maturity

Goto-Kakizaki (G-K) Rat

Diabetes-associated skeletal changes are well characterized in the G-K rat compared to other diabetic rodent models. The G-K rat was developed by selective inbreeding of the Wistar rats with elevated blood glucose to create a spontaneous diabetic rat model (154). These rats maintain glucose levels below 200 mg/dl through about 14–15 wks of age or just as skeletal maturity is approaching (155,156). Plasma glucose levels begin to increase beyond 200 mg/dl at approximately 15–16 wks of age and immediately begin to enter the diabetic range by 17–18 wks of age (157). Glucose challenge tests show that hyperglycemia in the G-K rat is driven by impaired insulin secretion and insulin resistance, but the primary cause is impaired insulin secretion (158). Although these rats become diabetic, they are not known to develop retinopathy (159), nephropathic changes occur only secondary to injury (160,161), and only moderate peripheral neuropathy symptoms appear to present (162,163).

The skeletal phenotype of the G-K rat includes a shorter femur and smaller lumbar vertebral height at 24 wks of age (164). G-K rats also have significantly lower femoral and lumbar BMD, but similar diaphyseal cross-sectional thickness and areas compared to controls (165). Histomorphometry of the proximal tibia corroborates BMD findings. The lower trabecular BV/TV in the G-K rat is due in part to decreased bone formation rate, combined with increased resorption, as evidenced by two-fold higher serum TRAP than controls (165). Finally, whole bone mechanical properties are decreased in G-K, with lower tibial (3-pt bending) and L5 (compression) peak loads, though the estimated elastic modulus of the tibia is similar to controls (165).

Polygenic/Lean/Post-Skeletal Maturity

Spontaneous Diabetic Torii (SDT) Rat

The SDT rat is a model of non-obese T2D that was developed by inbreeding of five polyuric and glycosuric Sprague Dawley rats. Diabetes develops in males but not females. The males have normal body weight, but fasting plasma glucose levels around 200 mg/dl at approximately 20 wks of age (60,62). After 24 weeks of age, glucose levels surpass 600 mg/dl. Insulin levels are lower in the SDT rat compared to the control due to low insulin secretion instead but insulin resistance is also suggested by fact that diabetes persists after insulin treatment or pancreas transplantation (62,166,167). Low insulin secretion is caused by beta-cell dysfunction (60). It continues until about 20 wks of age (62) when pancreatic islets begin to atrophy and insulin production decreases further. Among typical diabetic complications, the SDT rat develops some ocular changes but no retinal hemorrhaging as is in human diabetic retinopathy (60,168). Peripheral thermal neuropathy does develop with T2D (169). Human-like renal changes also develop with T2D, but some of the more overt and human-like changes occur (i.e., mesangial matrix expansion) at very advanced ages (beyond 32 wks) (170).

BMD and serum markers of bone formation (ALP and OCN) are similar between controls and SDT rats at 20 wks of age (171). However, by 30–36 wks of age, SDT rats have decreased femoral and tibial BMD, and reduced tibial three-point bending stiffness and peak load compared to controls (171,172). Trabecular histomorphometry indicates reduced bone formation rate, eroded surface, osteoblast and osteoclast numbers in SDT (171,172).

Polygenic/Obese/Pre-Skeletal Maturity

Zucker Diabetic Sprague Dawley Rat (ZDSD)

As stated above, the ZDF rat has an inactivating mutation in the leptin receptor that leads to hyperphagia and quick progression to frank diabetes. However, leptin receptor mutations are relatively rare in humans. In addition, leptin gene or receptor mutations impact BMD independently, confounding the effects of diabetes on bone. The ZDSD rat was bred to overcome this limitation of the Zucker Diabetic Fatty (ZDF fa/fa) rat. The ZDSD was created by breeding heterozygous ZDF rats with diet-induced obesity (DIO) rats (61). The result was a polygenic model of obesity and diabetes that develops diabetes at an older age than the ZDF rat. Obesity and diabetes develop spontaneously in both sexes and diabetes severity can be intensified with a high-fat diet (173,174). The disease develops between 15 and 17 weeks of age (61) in males, concurrent with the reported timing of rat skeletal maturity (56,61). The timing of frank diabetes in females may be later than males but evidence on the relative timing are still unclear (174). Approximately 50% of females develop diabetes compared to 100% in males. Glucose levels worsen with age in males, reaching 600–700 mg/dl (diabetes range) at 31–33 wks age (61,173). These worsening glycemic levels appear to develop, at least in part, due to low insulin secretion compared to controls (61). Morphological changes consistent with diabetic nephropathy have been reported (175), but retinopathy and neuropathy have not been reported for the ZDSD rat.

Adult male ZDSD rats have reduced femoral length, diameter, diaphyseal cortical thickness, and cortical vBMD (61). Femoral yield force, stiffness, and failure load, assessed by 3-pt bending tests, are lower in ZDSD as well. Reports of estimated cortical bone material properties are inconsistent (61). It is possible that the severity of the disease impacts these outcomes; femoral midshaft toughness post-yield toughness were significantly lower in animals fed a high-fat diet for ten weeks post-skeletal maturity while another study used a standard diet regimen and found no significant differences in these same parameters (61,176).

Lumbar vertebral size is normal, but vertebral BMD and trabecular BV/TV are lower in the ZDSD rats compared to controls. Accordingly, compressive yield force, stiffness, and energy-to-ultimate load are lower in ZDSD vertebrae compared to controls (61,176). At present, there is no information about whether these skeletal deficiencies are associated with altered bone remodeling rates, as there are no data on serum biomarkers or histomorphometry.

Polygenic/Obese/Post-Skeletal Maturity

Otsuka Long-Evans Tokashima Fatty Rat (OLETF)

The Otsuka Long-Evans Tokashima Fatty rat is a polygenic model of modest obesity and impaired glucose tolerance (59,177–179). Some reports suggest that the OLETF rat becomes pre-diabetic (210 mg/dl) or diabetic (279 mg/dl) at 40 wks of age or older (180,181) but these trends lack consistency across studies. Impaired glucose tolerance is clear after 24 wks of age since glucose levels minimally decrease even two hours after the introduction of a glucose bolus (59,177). Limited evidence suggests that even a long-term high-fat diet does not push fasting glucose levels into the diabetic range (>250 mg/dl) (182). As with several other rodent diabetic models, impaired glucose tolerance occurs in males and to a lesser extent in females (59). As it ages, the male rat experiences decreased insulin secretion due to degeneration of beta cells. Nephropathy progresses with age in a manner very similar to humans (59,183) as well as mild to moderate diabetic neuropathic symptoms (184–186). It is still unclear if retinopathy is a complication of late stage diabetes in the OLETF rat; preliminary reports are contradictory (187,188).

The skeletal phenotype of the OLETF model is very poorly characterized. Limited evidence indicates that it has high lumbar BMD compared to controls (Long-Evans Tokashima Otsuka rats) at all ages (189). In the tibial metaphysis, OLETF rats maintain higher BMD than controls until about 56 wks of age, when insulin secretion reaches a minimum (189). It is worth noting that OLETF rats also have normal tibial midshaft BMD until 56 wks of age, after which BMD falls below the controls (189).

There is limited information on serum markers of bone turnover, however one study indicates that OTELF rats have lower serum osteocalcin, normal TRAP and higher sclerostin levels compared to controls (179). The OLETF rat is difficult to acquire, only males are provided, and there are vendor restrictions on breeding these animals in a lab (pers. comm).

Conclusion

There is substantial and growing evidence that adults with type 2 diabetes (T2D) have an increased risk of fracture, despite obesity and normal to high bone mineral density. To date, the factors underlying increased fracture risk in T2D are poorly understood. Thus, there is a need for relevant animal models that can be used to gain insight into the mechanisms contributing to skeletal fragility in T2D.

Our review showed that although there are many rodent models of T2D, relatively few have undergone thorough assessments of bone metabolism and strength. Arguably, the key metabolic and skeletal traits that an animal model would ideally recapitulate include adult-onset, obesity-induced diabetes in both sexes, and exhibiting normal BMD. Yet, few of the current animal models achieve these goals (see Table 4), or if they do, the animal models themselves are difficult to acquire.

Table 4.

Summary of general skeletal characteristics of human adult onset T2D compared to commonly used rodent models

	Human T2D	Rat - ZDF (61,149,151)	Rat – ZDSD (61,174,176)
Underlying genetic cause	Polygenic	Mutation in leptin receptor	Polygenic
Age of onset	Post-maturity	Pre-maturity (early)	Pre- or at maturity
Sex predisposition	Men = Women	Male >> Female	Male > Female
Body weight	Increased	Increased	Increasedb
BMD	Normal to increased	Decreased	Decreased
Trabecular bone volume and | architecture	Normal to high	Low BV/TV	Low BV/TV
Cortical bone volume architecture	Possibly cortical porosity	Cortical thinning	Cortical thinning
Bone turnover	Decreased formation, normal or low resorption	Decreased or normal formation, normal to high resorption	no data
Marrow adiposity	Normal or increased	Increased	no data
Bone mechanical properties	↓ cortical tissue propertiesa	↓stiffness, yield & max load, post- yield energy	↓stiffness, yield & max load, apparent modulus
Fracture risk	Increased	(NA)	(NA)

(NA): not applicable

^aAs assessed by reference point indentation

^bBody weight begins to decrease after onset of frank diabetes

In particular, there are no mouse models that spontaneously develop diabetes after skeletal maturity. Among rats, there are no models that clearly develop obesity and diabetes after skeletal maturity. The OLETF rat develops moderate obesity and glucose intolerance after skeletal maturity, maintains normal BMD relative to control late in life, but does not appear to develop frank diabetes. The male ZDSD rat appears to develop T2D prior to or at skeletal maturity, suggesting that it enters a pre-diabetic state before skeletal maturity. Given this situation with spontaneous rodent models, diet-induced models offer a potential solution since the diet can be started after skeletal maturity. Yet, whereas the commonly used C57Bl/6 mouse strain is susceptible to diet-induced obesity, it is notably resistant to development of diabetes. Moreover, although the gerbil, Psammomys obesus, appears to replicate adult-onset, diet-induced diabetes well, its skeletal phenotype has not been adequately studied, and there are challenges related to acquisition, disease penetration, and the need for a specialized diet to maintain a non-diabetic state (190,191).

In the majority of rodent models, the onset of diabetes begins prior to skeletal maturity. Figure 1 graphically represents this for some mouse and rat models. This timing relative to skeletal maturity is generally associated with low bone mass, reduced bone formation rates, and weaker bones. Data on the impact of childhood diabetes on skeletal health are scant, but the global childhood obesity and diabetes epidemic reinforces the importance of these pre-skeletal maturity animal models of T2D. The distinction between childhood and adult onset diabetes may prove to be important because skeletal outcomes and treatment strategies for children and adults may be very different. Children predominantly experience modeling whereas adults predominantly experience remodeling, and the impact of diabetes could have different effects in these two situations. During growth type 2 diabetes may inhibit normal bone formation (192,193) lead to low bone mass relative to non-diabetic controls (194). Alternatively, adults with T2D appear to experience suppression of remodeling with normal BMD (2,4,37).

Figure 1.

Comparison of fasting glucose levels by age in (a) mouse and (b) rat models of T2D. Data are adapted from several sources: Mouse: (110,113,200), Rat: (61,146).

Our review also identified a need for models to study adult onset diabetes and skeletal fragility in females. Several of the spontaneous diabetic models, including the Tallyho and M16 mouse models, as well as the SDT and ZDF rat models, have a sex bias, such that either frank diabetes does not develop in females or disease penetration is low (59–61). A new exception appears to be the female ZDSD rat (174). More female models of diabetic skeletal fragility necessary to better understand the link between estrogen levels, adiposity, insulin resistance, and skeletal fragility.

Although obesity-associated diabetes is most common in Western societies, the SDT rat may be helpful in understanding the pathogenesis of diabetic skeletal fragility in some Asian populations where adult diabetic patients typically have a lean (low BMI) body type (4,195,196). Moreover, some Asian populations experience hypoinsulinemia and insulin resistance unlike U.S. and other Western populations (197,198). Population and diabetes trends now indicate that China has the greatest number of adults with T2D in the world (199), thus this area of investigation may gain additional importance.

A particular gap in knowledge is the impact of T2D on bone mechanical properties at both the whole bone and bone tissue levels (see Tables 2 and 3). Despite speculation about the possible negative effect of AGE’s on bone biomechanical properties, limited information is currently available on bone material properties in diabetic animal models, and the information available has yet to show a clear trend (61,68,157,176). Inconsistent trends can be attributed to different animal models, testing methods, and small sample sizes. Additional research in animal models and human bone tissue is critical to better understand the impact of diabetes on bone biomechanical properties.

In summary, animal models are likely to be helpful to improve understanding of the factors contributing to diabetes-related skeletal fragility. Unfortunately, current animal models are limited in their ability to recapitulate key metabolic features of adult-onset, obesity-related diabetes and their impact on bone metabolism, bone mass and strength. Moreover, the limitations of data from existing studies, where the onset of diabetes occurred prior to skeletal maturity, must be recognized. It is likely that no single animal model will recapitulate all of the features of diabetic skeletal fragility in humans. In particular, due to the lack of Haversian remodeling, it may be difficult to use rodent models to study the mechanisms underlying increased cortical porosity in postmenopausal women with T2D and fractures. Large animal models, or at least those with Haversian remodeling, may be needed to study this issue.

In conclusion, optimal management of patients with type 2 diabetes will require additional knowledge about the factors contributing to their increased skeletal fragility. Given the exponential growth in diabetes worldwide, our review has highlighted an urgent need to develop new animal models that better reflect the clinical characteristics of diabetic skeletal fragility, namely where the onset of diabetes is after skeletal maturity, occurs in both sexes, and leads to normal bone mass but reduced bone strength.