Altered bone microarchitecture in a type 1 diabetes mouse model Ins2Akita

Type 1 diabetes mellitus (T1DM) has been associated to several cartilage and bone alterations including growth retardation, increased fracture risk, and bone loss. To determine the effect of long term diabetes on bone we used adult and aging Ins2 Akita mice that developed T1DM around 3–4 weeks after birth. Both Ins2 Akita and wild‐type (WT) mice were analyzed at 4, 6, and 12 months to assess bone parameters such as femur length, growth plate thickness and number of mature and preapoptotic chondrocytes. In addition, bone microarchitecture of the cortical and trabecular regions was measured by microcomputed tomography and gene expression of Adamst‐5, Col2, Igf1, Runx2, Acp5, and Oc was quantified by quantitative real‐time polymerase chain reaction. Ins2 Akita mice showed a decreased longitudinal growth of the femur that was related to decreased growth plate thickness, lower number of chondrocytes and to a higher number of preapoptotic cells. These changes were associated with higher expression of Adamst‐5, suggesting higher cartilage degradation, and with low expression levels of Igf1 and Col2 that reflect the decreased growth ability of diabetic mice. Ins2 Akita bone morphology was characterized by low cortical bone area (Ct.Ar) but higher trabecular bone volume (BV/TV) and expression analysis showed a downregulation of bone markers Acp5, Oc, and Runx2. Serum levels of insulin and leptin were found to be reduced at all‐time points Ins2 Akita. We suggest that Ins2 Akita mice bone phenotype is caused by lower bone formation and even lower bone resorption due to insulin deficiency and to a possible relation with low leptin signaling.

different factors like higher glucose serum concentration and lower insulin secretion by the β cells, inflammation and altered gene expression. Advanced glycation end products (AGEs) are proteins or lipids that are formed in hyperglycemic environments. Since their cumulative effects increase with age, they represent a key player in vascular disease associated to diabetes (Goldin, Beckman, Schmidt, & Creager, 2006). AGEs are involved in an increase in inflammatory activity and a decrease in bone formation due to osteoblastic apoptosis and decreased osteoblast proliferation (Gangoiti, Anbinder, Cortizo, & McCarthy, 2013) or higher osteoclastic activation (Sanguineti, Puddu, Mach, Montecucco, & Viviani, 2014), as well as chondrocyte apoptosis in cartilage (Tsai et al., 2013). The receptor for AGEs (RAGE) is assumed to be the molecular intervenient that activates the pathways leading to oxidative stress and inflammation (Ramasamy, Yan, & Schmidt, 2012) including in bone since osteoblasts, osteoclasts and chondrocytes express RAGE (Mercer, Ahmed, Etcheverry, Vasta, & Cortizo, 2007;Nah et al., 2007).
Hypoinsulinemia present in T1DM can also affect bone metabolism, since insulin signaling in osteoblasts was found to regulate bone resorption by activating osteoclastic activity (Ferron, Wei, & Yoshizawa, 2010), releasing undercarboxylated osteocalcin to the blood stream, which in turn affects glucose homeostasis by signaling insulin secretion in β cells and other insulin sensitive tissues . This relationship between bone and insulin was demonstrated when Ob-IR mice, lacking the insulin receptor (IR) only in osteoblasts, became glucose intolerant (Ferron et al., 2010).
Both T1DM patients and mice models face a rapid weight loss during the onset of the disease (Coe et al., 2012;), that persists if not treated, creating a state resembling an accelerated fast that results in loss of fat and proteins. Weight loss has been associated with low bone mass , but interestingly only a decrease in levels of fat mass were found to be correlated with decreased bone mineral density (BMD) and not lean mass or total body weight (Fogelholm, Sievänen, Kukkonen-Harjula, & Pasanen, 2001). This close relationship between fat and bone seems to be explained by the fact that adipocytes secrete leptin.
Accordingly, both the ob/ob leptin and the db/db leptin receptor mutant mice have impaired bone formation, exhibiting a normal or decreased cortical bone volume (BV/TV) although presenting a higher trabecular bone volume (BV/TV; Ducy et al., 2000;Turner et al., 2013). It was postulated that leptin binding to its receptors in the hypothalamus increases the expression of noradrenaline activating β2-adrenergic receptors pathway in osteoblasts, inhibiting bone formation and increasing the expression of receptor activator of nuclear factor κΒ ligand, promoting the differentiation and proliferation of osteoclasts (Ducy et al., 2000;Elefteriou et al., 2005). These findings were supported by the results of β2-adrenergic receptor KO mice (Adrb2 −/− ), that exhibit an increase in trabecular bone at the age of 6 months (Elefteriou et al., 2005). It has been proposed that peripherally, leptin promotes osteoblast proliferation through leptin receptor signaling (Cornish et al., 2002) and more recently, Turner et al. (2013) proposed that peripheral leptin induces bone formation and resorption, this representing the main route of action of leptin in bone. Food intake is correlated with high levels of leptin and fasting periods are associated with low levels of leptin, as previously observed in fasting mice and in anorexia nervosa (Devlin et al., 2010;Soyka, Grinspoon, Levitsky, Herzog, & Klibanski, 1999). Devlin et al. (2010) in their experiments with mice under caloric restriction (CR), from 3 to 12 weeks of age, not only correlated leptin levels with CR but also with low cancellous BV/TV and low cortical area, assuming that CR in juvenile mice under a fast period of growth lead to bone loss. But unexpected results were observed in 6 months mice after a period of 10 weeks under CR (Hamrick, Ding, Ponnala, Ferrari, & Isales, 2008), which presented low cortical mass, but higher trabecular BV/TV in the vertebra and unchanged trabecular BV/TV in the femur. In our study we hypothesize that inflammation, together with insulin deficiency and a possible decrease in leptin signaling, could be the principal causes involved in the cartilage and bone phenotypes observed in Ins2 Akita .

| Mouse models
Five male wild-type (WT) C57BL/6 and five male heterozygous Ins2 Akita (C57BL/6 background) were sampled at each of the stages analyzed, 4, 6, and 12 months of age (total of 30 mice) and were used to perform the experimental procedures. Diabetes was monitored by blood glucose measurements using a glucose assay kit (Free Style Precision; Abbott Laboratories, Chicago, IL) and only Ins2 Akita mice with glucose values >300 mg/dl were used in this experiment. All animal manipulations were conducted in accordance with principles and procedures following the guidelines from the Federation of Laboratory Animal Science Associations (FELASA). Mutant and wildtype mice were originally purchased from Jackson Laboratory (Bar Harbor, Maine) and colonies established in the local bioterium at the University of Algarve. All animals were kept on a light/dark (12 hr/ 12 hr) cycle at 23°C, and received food (standard lab chow) and water ad libitum.

| Total RNA isolation
Left femur and tibia were isolated and cleaned from adhering tissues, the bone marrow was flushed out with phosphate buffered saline (PBS) and the bone was snap-frozen in liquid nitrogen. Frozen bones were crushed using a mortar and pestle under liquid nitrogen and RNA extracted with the Isol-RNA Lysis Reagent 5 PRIME ® (Hilden, Deutschland) according to manufacturer's protocol. RNA integrity was verified using Experion TM RNA Analysis Kit (Bio-Rad, Hercules, CA).

| 9339
Carlsbad, CA) according to the manufacturer's protocol. qRT-PCR was performed using the iQ™ SYBR ® Green Supermix (Life Technologies, Frederick, MD) and specific primers on an CFX96 Touch™ Real-Time PCR Detection System (Bio-Rad) for 40 cycles, each with 15 s for annealing and 30 s for amplification, followed by a melt curve analysis, as described (Technologies, 2011). All gene expression data were normalized against hypoxanthine phosphoribosyltransferase 1 (Hrpt1), and relative quantification calculated according to the −ΔΔ 2 C t method as previously described (Pfaffl, 2004).

| Serum measurements
Blood serum from three WT and Ins2 Akita at 4, 6, and 12 months was collected and stored at −80°C. Leptin was measured using a Novex

| Statistical analysis
All statistical analyses were performed using Stata Statistical Software (Stata Corp LLC, College Station, TX). The data was evaluated using one-way analysis of variance followed by Bonferroni multiple comparisons test, with p < 0.05 considered statistically significant.
Results are presented as means ± standard deviation (SD).

| T1DM reduces femur length and body weight in Ins2 Akita
Diabetes was confirmed by increased glucose concentrations observed at all-time points in Ins2 Akita mice when compared with WT, with an increase of 271%, 306%, and 356% at 4, 6, and 12 months, respectively, compared with age matched controls ( Figure 1).
Diabetic mice also presented a significant decrease in femur length

| Growth plate thickness is lower in Ins2 Akita
Since potential alterations in growth plate structure are known to impair longitudinal bone growth, we investigated if the growth plate of Ins2 Akita could be affected. Growth plate measurements ( Figure 3a) showed that thickness was reduced in Ins2 Akita at 4 and 6 months, but at 12 months no significant differences could be observed compared with the WT counterparts ( Figure 3b).
Taking into account that bone lengthening depends on the proliferation of chondrocytes in the growth plate, we determined the total number of proliferative chondrocytes and found a significant reduction in this number at 4 and 6 months in Ins2 Akita 3.3 | Ins2 Akita have lower cortical area and higher trabecular bone volume at 4, 6 at 12 months Total area (Tt.Ar) of cortical bone in the femur of Ins2 Akita was found to be reduced by 32% at 4 months, 16% at 6 months, and 25% at 12 months of age (Table 1 and Figure 5a). These differences were significant for all three age groups (p < 0.05; Table 1). The reduction was mainly due to a substantial decrease in cortical area (Ct.Ar), of 53% at 4 months, 35% at 6 months and 25% at 12 months, with all results being highly significant compared with WT controls (p < 0.001; Table 1 and Figure 5b).
This thinning of cortical bone observed in the diabetic mice was confirmed by a decrease in the cortical area fraction (Ct.Ar/Tt. Ar), cortical thickness (Ct.Th), and periosteal perimeter (Ps.Pm; p < 0.05; Table 1). No significant differences were observed for the marrow area (Ma.Ar). A significant reduction could be found in the endocortical perimeter (Ec.Pm; p < 0.05) at 12 months (Table 1)  Differences in bone volume relative to trabecular volume (BV/ TV) in Ins2 Akita were found to be highly significant at 4 months with an increase of 45% (p < 0.001) at 4 months, of 46% (p < 0.05) at 6 months and of 30% (p > 0.05) at 12 months (Table 1 and at 6 months (p < 0.05) and 43% at 12 months, and not due to the size of the trabeculae, since no differences were observed in the specific bone surface (BS/BV) or in the trabecular thickness (Tb. Th; Table 1). In Ins2 Akita the high Tb.N led to a highly significant (p < 0.001) reduction in trabecular separation (Tb.Sp) parameters in all three time points analyzed (Table 1)

| Expression of cartilage and bone marker genes is altered in Ins2 Akita
To determine the mechanisms leading to alterations in the cartilage of Ins2 Akita , we examined the expression levels of Adamst-5, which is involved in cleavage of proteoglycans, and Col2, the most abundant protein in cartilage. Adamst-5 was found to be overexpressed at alltime points in Ins2 Akita , being highly expressed at 4 and 12 months (p < 0.001) and also significantly upregulated at 6 months (p < 0.05; Figure 8a). Col2 expression was found to be downregulated at both 4 and 6 months (p < 0.05) compared with WT ( Figure 8b). Igf1 gene expression levels were found to be downregulated at 6 months (p < 0.05) and strongly downregulated at 4 and 12 months (p < 0.001; Figure 8c). Oc was found to be downregulated at 4 months (p < 0.05; Figure 8d). Expression levels of Runx2 (Figure 8e), the main transcription factor involved in osteoblast differentiation, were significantly downregulated at 4 and 6 months (p < 0.05) in Ins2 Akita and finally the osteoclast marker Acp5 was also found to be significantly downregulated at 4 (p < 0.01) and 6 months (p < 0.05; Figure 8f).

| Serum concentrations of insulin and leptin is reduced in Ins2 Akita
Blood serum concentrations of insulin and leptin was determined by enzyme-linked immunosorbent assay (ELISA) and in both cases were found to be significantly reduced when compared with WT at 4, 6, and 12 months (p < 0.001; Figure 9a,b).
F I G U R E 2 Ins2 Akita presents shorter femurs. (a) X-ray analysis of Ins2 Akita and wild-type (WT) mice femurs at 4, 6, and 12 months; (b) Ins2 Akita femur length is significantly smaller than WT at 4, 6, and 12 months, demonstrating that type 1 diabetes mellitus is related to growth retardation. (c) Body weight of Ins2 Akita is significantly lower compared with WT at 4, 6 and 12 months. Five animals per group and time point were evaluated. *p < 0.05, **p < 0.001. Error bars represent SD [Color figure can be viewed at wileyonlinelibrary.com] 4 | DISCUSSION T1DM has been associated to bone growth retardation in puberty (Donaghue, 2003) and increased risk of fracture throughout life, leading to higher morbidity and mortality (Weber, Haynes, Leonard, Willi, & Denburg, 2015). Higher bone porosity and smaller cortical area are the principal causes for the observed decrease in biomechanical properties, as previously reported for type 2 diabetic postmenopausal women (Patsch et al., 2013). In the current study, growth retardation could also be observed in the T1DM mice model ). An increase in tumor necrosis factor α (TNF-α) is known to affect bone environment  and has been associated with upregulation of aggrecanase 5 (Adamst-5), a metalloproteinase exerting a potent effect on cartilage matrix degradation (Illien-Junger et al., 2013). Accordingly, this enzyme was found to be highly expressed in our study, likely contributing to higher cartilage degradation. Our results showed low levels of Igf-1 expression in Ins2 Akita at all-time points. Lower circulating Igf-1 concentrations have been associated with reduced linear growth (Yakar et al., 2002), higher cartilage degradation, and lower chondrocytic and osteoblastic proliferation (Kasukawa, Miyakoshi, & Mohan, 2004). Serum Igf-1 was also found to be lower in CR mice (Devlin et al., 2010; caused by impaired growth hormone signaling (LeRoith & Yakar, 2007). These results suggest that a decrease in Igf-1 signaling might be involved in the reduction of bone quality parameters observed in our diabetic subjects.
Lower insulin signaling in adipocytes and weight loss in diabetes leads to low expression of leptin (Martin & McCabe, 2007) and constitutes what Ins2 Akita may have in common with previous models that could explain these similarities is leptin deficiency. This result is similar to what was observed in this study with low levels of insulin detected in Ins2 Akita mice after the onset of disease and the reduction of insulin signaling can also explain the reduced levels of leptin observed at the same ages analyzed. To explain the mosaic phenotype,  suggested that under caloric restriction there is a leptin deficiency and an increased neuropeptide Y signaling leading to reduced cortical bone. Baldock et al. (2006) reported an increase in cortical bone volume in Y2 receptor KO mice, months and, more recently, at 4 months (Elefteriou et al., 2005;Pierroz et al., 2012). In our study we could detect this increase in trabecular bone in Ins2 Akita starting at 4 months. It was shown by Ducy et al. (2000) that ob/ob and db/db mutant mice at 6 months had higher trabecular BV/TV both in vertebrae and in tibia. To explain the high trabecular volume and low cortical bone volume, it has been shown that leptin have a neuroendocrine role increasing the expression of osteogenic markers related to bone formation, but also to stimulate bone resorption (Bartell et al., 2011;Hamrick et al., 2004) or by suggesting a higher significance of the stimulatory effect of leptin in bone peripherally. Turner et al. (2013) have proposed that leptin can influence bone by acting centrally and peripherally, and in both cases leptin induces bone formation and resorption, concluding that regulation was predominantly made by direct signaling on both the osteoblastic and osteoclastic lineages.
Lower Oc and even lower cross-linked C-telopeptide serum levels in leptin mutant ob/ob and in the leptin receptor mutant db/db mice was associated with low bone formation and low bone resorption (Turner et al., 2013). These conclusions led to the assumption that higher bone volume in the trabecular bone of the vertebrae was due to lower bone formation but an even higher reduction in bone resorption. Turner et al. (2013) proposed an interesting model to in Ins2 Akita were always significantly higher, the number of osteoclasts and Acp5 expression was reduced at 4 and 6 months while the expression of genes associated with bone formation (Oc and Runx2) showed to be downregulated, particularly at 4 months, when we could detect higher histomorphometric differences in trabecular and cortical bone. It has also been shown by Kalra, Dube, and Iwaniec (2009) that 10 weeks old Akita mice had significantly lower plasma Oc than WT mice, confirming a lower osteoblastic activity. Other explanation for the presumable lower bone formation and resorption rate expressed by our results, is the fact that insulin signaling in osteoblasts has been associated to higher osteoblast and osteoclast activity promoting both bone formation and resorption . Hyperglycemia has been associated to lower bone quality, especially by the role of AGEs that have been shown to reduce osteoblastic differentiation and by increasing osteoclast bone resorption. These findings are supported by work with KO mice for the receptor for AGEs (RAGE), that presented higher bone volume and lower bone resorption (Zhou, Foster, Zhou, Cowin & Xian, 2006).
Although the possible higher signaling of AGEs in osteoblasts resembles the lower bone volume observed in our study, osteoclast activation by RAGE conflicts with our data and with the majority of reports with type 1 diabetic models ) that suggests lower bone resorption. Nevertheless, AGEs are thought to be preponderant in reducing the biomechanical properties of bone, since they accumulate in bone matrix, reducing bone strength and increasing fracture risk (Yamamoto, Yamaguchi, Yamauchi, Yano, & Sugimoto, 2008).
High bone marrow adiposity has been associated to reduced bone formation (Devlin et al., 2010), due to the fact that adipogenesis and osteoblastogenesis are derived from a common mesenchymal precursor and selection of adipose lineage could lead to reduced number of osteoblasts although this hypothesis as not yet been confirmed ). It has been proposed that marrow adipose tissue may act physiologically to provide an expandable/contractible fat depot for sustaining optimal hematopoiesis (Turner, Martin, & Iwaniec, 2018). Also inflammation in bone environment has been pointed as a possible cause for reduced bone formation, when the MC3T3 osteoblastic cell line was exposed to bone marrow from diabetic mice it resulted in increased osteoblast death, but when cocultured with TNF-α neutralizing antibodies the cell death response was reduced (Coe et al., 2011).
Reduced bone formation in T1DM seems to have multifactorial explanations, but reduced bone resorption can be explained, in part, by the reduced insulin and leptin signaling in osteoblasts and/or osteoclasts, as previously reported. Like in previous reports (Jun, Ma, Pyla, & Segar, 2012;Naito et al., 2011;Schoeller et al., 2014), Ins2 Akita in our study showed to be insulin and leptin deficient, and this double disorder may explain why diabetic mutants presented such marked differences, where in the trabecular region of the Ins2 Akita at 4 months bone volume was almost two times higher. Although