Differential Bone Metabolism and Protein Expression in Mice Fed a High-Fat diet Versus Pre-Hibernation Fattening in Daurian Ground Squirrels: A Comparison Between Pathological and Healthy Obesity


 Background

This study compared effects on bone metabolism and morphology of pathological obesity induced by excessive fat intake in a non-hibernator (mice) versus healthy obesity due to pre-hibernation fattening in a hibernator (ground squirrels).
Methods

Kunming mice were fed a high-fat diet for 3 months to provide a model of pathological obesity (OB group). Daurian ground squirrels fattened naturally in their pre-hibernation season (PRE group) were used as a healthy obesity model. Body weight and adipose tissue wet weight were measured. Micro-computed tomography and three-point bending tests were used to determine the microstructure and mechanical properties of bone. Western blots were used to analyze protein expression levels related to bone formation (RunX2, OCN, ALP), bone resorption (RANKL, Cathepsin K, MMP9) and Wnt signaling (P-β-catenin, GSK-3β).
Results

Micro-CT showed that there was no obvious bone loss in the OB group, compared with controls, whereas bone formation in the PRE group was enhanced, compared to summer-active squirrels. Results of three-point bending tests showed that stiffness of the femur in OB mice was significantly enhanced, whereas the mechanical properties of bone in the PRE group did not change. Analysis of bone protein expression showed significantly increased expression levels of ALP, OCN, RANKL, MMP9, Cathepsin K, GSK-3β and P-β-catenin in OB mice, but RunX2 expression did not change. By contrast, PRE ground squirrels, showed significantly increased expression of most proteins except for OCN that decreased significantly, and P-β-catenin that did not change.
Conclusion

For non-hibernating mice, moderate obesity had a certain protective effect on bones, demonstrating two-way regulation, enhancing both bone loss and bone formation, so that bone metabolism was at a higher level of metabolic balance. For pre-hibernating ground squirrels, the healthy obesity acquired before hibernation had a positive effect on the microstructure of bones and also enhanced the expression levels of proteins related to bone formation, bone resorption and Wnt signaling.


Abstract Background
This study compared effects on bone metabolism and morphology of pathological obesity induced by excessive fat intake in a non-hibernator (mice) versus healthy obesity due to pre-hibernation fattening in a hibernator (ground squirrels).

Methods
Kunming mice were fed a high-fat diet for 3 months to provide a model of pathological obesity (OB group). Daurian ground squirrels fattened naturally in their pre-hibernation season (PRE group) were used as a healthy obesity model. Body weight and adipose tissue wet weight were measured. Micro-computed tomography and three-point bending tests were used to determine the microstructure and mechanical properties of bone. Western blots were used to analyze protein expression levels related to bone formation (RunX2, OCN, ALP), bone resorption (RANKL, Cathepsin K, MMP9) and Wnt signaling (P-βcatenin, GSK-3β).

Results
Micro-CT showed that there was no obvious bone loss in the OB group, compared with controls, whereas bone formation in the PRE group was enhanced, compared to summer-active squirrels. Results of threepoint bending tests showed that stiffness of the femur in OB mice was signi cantly enhanced, whereas the mechanical properties of bone in the PRE group did not change. Analysis of bone protein expression showed signi cantly increased expression levels of ALP, OCN, RANKL, MMP9, Cathepsin K, GSK-3β and Pβ-catenin in OB mice, but RunX2 expression did not change. By contrast, PRE ground squirrels, showed signi cantly increased expression of most proteins except for OCN that decreased signi cantly, and P-βcatenin that did not change.

Conclusion
For non-hibernating mice, moderate obesity had a certain protective effect on bones, demonstrating twoway regulation, enhancing both bone loss and bone formation, so that bone metabolism was at a higher level of metabolic balance. For pre-hibernating ground squirrels, the healthy obesity acquired before hibernation had a positive effect on the microstructure of bones and also enhanced the expression levels of proteins related to bone formation, bone resorption and Wnt signaling.
Human obesity has become a major epidemic around the world [1]. Obesity caused by excessive fat intake causes a variety of disease states, including bone structure destruction and bone loss [2,3]. Previous studies with model species have shown that nutritional obesity will lead to massive bone loss, decreased size-independent mechanical properties, destruction of bone microstructure, abnormal bone metabolism and so on [4][5][6][7]. For example, the hind limb bones of male 6-week-old C57BL/6J mice (Mus musculus) fed a high-fat diet for 17 weeks showed a decrease in bone mass, bone density and bone strength [8]. In both 3 week-old and 15-week-old male C57BL/6 mice fed a high-fat diet for 16 weeks, obesity was accompanied by massive bone loss and bone microstructure destruction [9]. In addition, nutritional obesity may also lead to a decrease of mechanical properties. Studies showed that male C57BL/6 mice fed a high-fat diet for 12 weeks had reduced femoral mineral density and bone ductility [10]. A similar study with male C57BL/6 mice fed a high-fat diet also showed a decrease in bone mineral density [11]. Other studies con rm that nutritional obesity induced by a high-fat diet usually inhibits bone formation, but bone resorption was enhanced or unchanged [12]. However, the speci c mechanisms of this bone loss are not yet fully understood. Therefore, an in-depth understanding of the effects of obesity on the skeletal system may help prevent obesity-induced bone loss.
Different from nutritional obesity, a natural model of fattening is the seasonal acquisition of huge fat reserves by some mammals prior to winter hibernation. Fat storage begins weeks before hibernation, with a period of hyperphagia that greatly increases body mass [13]. For example, Arctic ground squirrels (Spermophilus parryii) and marmots (Marmota aviventris) often gain approximately double their body mass [14,15]. However, this kind of obesity is not accompanied by harmful diseases, such as type 2 diabetes, hyperglycemia and hyperlipidemia that are common in human obesity [16]. Therefore, the obesity caused by fattening before hibernation is called healthy obesity [17]. Although the weight gain of these hibernating animals is much greater than that of nutritionally obese animals, they do not experience osteoporosis, and their bone strength and bone microstructure do not change signi cantly [18].
Bone metabolism is regulated by bone formation and bone resorption [19]. Osteoblasts are the main cells involved in bone formation [20]. The differentiation of osteoblasts is regulated by a variety of factors, including Runt-related transcription factor 2 (RunX2), osteocalcin (OCN), and bone-derived alkaline phosphatase (ALP) [21]. RunX2 is a crucial transcription factor for osteoblast differentiation and plays a vital role in bone development [22,23]. RunX2 not only induces the activation of the OCN promoter to accelerate bone mineralization, but also interacts with Wnt signaling to promote osteoblast differentiation [24,25]. When the body is extremely obese, RunX2 is down-regulated. Studies have shown that the RunX2 mRNA levels decreased in 5-week-old male Sprague Dawley (SD) rats (Rattus norvegicus) fed with a high-fat diet for 12 weeks [26]. OCN is one of the signs of bone formation and can promote bone mineralization [27]. Studies have shown that the expression of OCN decreased in Col1a1 Jrt/+ mice fed a high-fat diet for 26 weeks [28]. In addition, ALP is a key enzyme marker of bone formation due to its action in regulating the process of biomineralization [29]. Previous studies have shown that the expression level of ALP decreased in 6-week-old male C57BL/6 mice fed a high-fat diet for 6 months [30]. Therefore, the expression levels of bone formation marker proteins (RunX2, OCN and ALP) can be useful markers that can re ect the status of bone formation metabolism.
Osteoclasts play an important role in bone resorption [21]. Nuclear factor kappa B receptor activator ligand (RANKL), is a membrane-bound protein present in osteoclasts and bone cells and is a receptor activator necessary for the differentiation of osteoclast precursors into osteoclasts [31]. Studies have shown that the expression level of RANKL was up-regulated in the left femurs of 5-week-old male Wistar rats fed a high-fat diet for 6 weeks after 4 weeks of caloric restriction [32]. Cathepsin K is also necessary for normal bone resorption. It can degrade the tissue matrix and directly regulate the expression of osteoclast bone resorption factors, including cytokines, hormones and nuclear transcription factors [33]. Previous studies have shown that Cathepsin K was up-regulated in 3-week-old female Wistar rats fed a high-fat diet for 5 weeks [34]. Metallopeptidases (MMP) are a protein family composed of zinc-dependent endopeptidases that regulate tissue remodeling under physiological and pathological conditions [35]. Among them, MMP9, also known as gelatinase B or 92-kDa type IV collagenase, is responsible for degrading the extracellular matrix [35]. The up-regulation of MMP9 may trigger or aggravate the hydrolysis of proteins in the bone matrix and promote bone resorption. Under obesity and hyperglycemia, the expression level of MMP9 is signi cantly increased [36]. Therefore, the expression levels of bone lossrelated proteins (RANKL, Cathepsin K and MMP9) re ect bone loss metabolism.
In recent years, more and more studies have shown that bone formation and bone resorption are regulated by Wnt signaling [37]. Wnt/β-catenin signaling plays an important role in the development and functional regulation of osteoblasts[38], activation of the Wnt pathway promoting the proliferation and differentiation of osteoblasts [39]. When Wnt/β-catenin signal transduction is disturbed, a variety of bone diseases such as osteoporosis can occur [40]. Studies have shown that decreased activity of Wnt/βcatenin signaling leads to decreased bone formation in male rats fed a high-fat diet for 4 weeks [41].
When the Wnt ligand is missing, glycogen synthase kinase 3β (GSK-3β) phosphorylates β-catenin, leading to a rapid degradation of P-β-catenin by the proteasome [42]. After the Wnt receptor is activated, GSK-3β is inhibited, leading to accumulation of non-phosphorylated β-catenin and activating target genes involved in regulating the proliferation and differentiation of bone marrow stromal cells [43]. It has been shown that β-catenin deletion inhibits osteoblast differentiation [44]. In addition, Wnt signaling can indirectly affect the function of osteoclasts by regulating the expression of bone resorption-related proteins [45].
Wnt signaling causes less bone resorption by down-regulating RANKL in the osteoclasts [46]. However, research on Wnt signaling in the fattening stage before hibernation is currently imperfect.
Based on our previous research, we chose to compare a healthy obesity model, the Daurian ground squirrel (Spermophilus dauricus) that naturally fattens before hibernation but does not show associated muscle atrophy, with an obesity mouse model induced by a high-fat diet that does show atrophy [13]. We aimed to determine if the skeletal system of pre-hibernation ground squirrels has a special mechanism to avoid the bone loss that occurs in the high-fat obesity model of mice. Little is known to date about the changes in bone of fattening animals before hibernation, and the related mechanisms are not clear. Therefore, an in-depth understanding of the bone state in the fattening stage before hibernation and a comparison of the differential regulatory mechanisms of pathological obesity versus healthy obesity in bone formation and bone resorption are of signi cance for gaining a greater understanding of the mechanisms that can prevent bone loss. We propose the following hypothesis: there are differences in bone metabolism between the two types of obesity models, which are partly achieved by regulating the expression levels of proteins related to bone formation, bone resorption, and Wnt signaling. We used Kunming mice fed a high-fat diet for 3 months as a pathological model for nutritional obesity and Daurian ground squirrels fattened before hibernation as a healthy obesity model. Hind limb bones were used to compare bone microstructure, mechanical properties, expression levels of bone formation-related proteins (RunX2, OCN and ALP), bone resorption-related proteins (RANKL, Cathepsin K and MMP9) and Wnt signaling proteins (P-β-catenin and GSK-3β). The data further clari es the differential regulation and mechanisms of bone remodeling occurring between models of pathological obesity and healthy obesity. mice were kept in plastic cages in the animal room and provided with food and water ad libitum. The animal room was maintained at a temperature range of 18-25°C, and lighting was changed daily to coincide with local sunrise and sunset. After a week of normal diet feeding, the mice were randomly divided into two groups (n = 6): CON: Control mice fed a normal diet for 3 months; and OB: Obesity mice fed a high-fat diet for 3 months. The normal feed and high-fat diet feed were purchased from Chengdu Dashuo Experimental Animal Company and the compositions of both are shown in Table 1. After the 3 months dietary intervention, all mice were sacri ced. Ground squirrels were obtained as previously described by our laboratory [47]. Brie y, twelve Daurian ground squirrels of both sexes were caught from the Weinan region in Shaanxi Province of China. Ground squirrels were kept in plastic cages in the animal room and provided with food and water ad libitum. The animal room was maintained at a temperature range of 18-25°C, and lighting was changed daily to coincide with local sunrise and sunset. Ground squirrels were divided into two groups (n = 6): summer active (SA) controls that were captured and sacri ced at the end of June, and the pre-hibernation (PRE) group that were captured and sacri ced at the end of September, after natural fattening. Both groups of ground squirrels acquired natural foods before being caught. After returning to the lab, they were fed rat chow with the same composition as the normal diet in Table 1A. In order to ensure the nutritional requirements during the fattening period, appropriate amounts of high-fat and high-protein nuts (such as peanuts) were added to the PRE ground squirrels.

Sample collection
After body weight was recorded, mice were anesthetized with 2 mg/kg sodium pentobarbital intraperitoneally, and ground squirrels with a 90 mg/kg dose. Then, the femurs and tibias from both legs were carefully dissected free of associated connective tissue, weighed quickly, and immediately placed in sealed containers with lactated 70% absolute ethanol, followed by freezing in liquid nitrogen and storage at -70℃. We also sampled adipose tissues, including mesenteric adipose, perirenal adipose and back scapula subcutaneous adipose. Similarly, adipose tissues were weighed, then frozen in liquid nitrogen and storage at -70°C. Animals were euthanized by an overdose injection of sodium pentobarbital after sampling.

Biomechanical testing
Mechanical properties of the femurs were determined by loading the left femurs to failure in a 3-point bending test, exactly as described previously with female mice [48]. Three parameters -ultimate bearing capacity, stiffness and ultimate bending energy -were determined as described previously [49]. Stiffness was de ned as the slope of the linear region of the pre-yield load displacement curve, and the yield point was de ned as the point where the load-displacement curve intersected with a regression line that was 10% lower than that used to de ne stiffness. Maximum load was de ned as the load at which the bone catastrophically failed.

Micro-computed tomography (micro-CT)
The right femurs and tibias of mice and ground squirrels were scanned using a micro-CT scanner (L-SP, GE, USA). Then the images were analyzed by GEHE Microview V2.

Western blots
Methods were those described previously by our lab [51]. Brie y, total protein was extracted from frozen femurs of mice and ground squirrels by homogenization and put into a sample buffer (pH 6.8, 100 mM Tris, 4% SDS, 5% glycerol, 5% 2-β-mercaptoethanol, and bromophenol blue). Then, the bone protein extracts were separated by SDS-PAGE (10% Laemmli gels with an acrylamide/bisacrylamide ratio of 29:1 for ALP, P-β-catenin, Cathepsin K, GSK-3β, MMP9, OCN, RANKL and RunX2). After electrophoresis, total proteins bands were visualized by putting the gel on the UV transilluminator and irradiating the gel for 2 min, and a Syngene G:BOX system (Syngene, Frederick, MD) was used to take photographs of the gel. Proteins in the gels were then transferred to PVDF membranes (0.45 µm) using a Bio-Rad semi-dry transfer apparatus. Membranes were blocked with 5% skim milk dissolved in TBST (10 mM Tris-HCl, 150 mM NaCl, 0.05% Tween-20.

Statistical analyses
An independent-samples t-test was used to determine the signi cant differences between the OB and the CON mice, or the PRE and the SA ground squirrels. All data were analyzed using SPSS 24 and expressed as means ± SD. A value of P < 0.05 was considered to be statistically signi cant.

Body weight
The composition of the normal and high fat diets fed to mice is shown in Table 1. The fat calorie percentage of the high fat diet was 61%. Table 2 shows that after feeding with high-fat diet versus normal diet for 3 months, the OB group had gained 10.6% higher body weight than the CON group (OB: 52.0 ± 0.7 g vs. CON: 47.0 ± 0.8 g, P < 0.05). In ground squirrels, the body weight of the PRE group was also signi cantly higher by 62.5% as compared with the SA group (PRE: 347.7 ± 10.8 g, vs. SA: 214.0 ± 6.4 g, P < 0.05), a mean rise of 134 g per animal over the pre-hibernation fattening period.

Adipose tissue wet weight
As shown in Table 3, compared with the CON group of mice, the mesenteric adipose wet weight in the OB group was signi cantly increased by 2.3-fold from 0.89 g to 2.01 g (P < 0.05), whereas the perirenal adipose wet weight increased signi cantly by 2.3-fold from 0.28 g to 0.63 g (P < 0.05) in mice. Compared with the SA group of ground squirrels, the mesenteric adipose wet weight in the PRE group was signi cantly increased by 22.5-fold from 0.87 g to 19.54 g (P < 0.05), the perirenal adipose wet weight also increased signi cantly by 8.2-fold from 0.39 g to 3.21 g (P < 0.05), and the subcutaneous adipose wet weight rose signi cantly by 37.9-fold from 0.35 g to 13.25 g (P < 0.05) in ground squirrels.

Femoral bone structure and function
Representative micro-CT images of distal femur of mouse CON and OB groups and ground squirrel SA and PRE groups are shown in Fig. 1. The images indicate no differences in the structure of the mouse femur between the CON and OB groups and substantial differences in the structure of the ground squirrel femur between the SA and PRE groups. This is backed up by quanti cation of multiple femoral bone parameters presented in Table 4. These quanti ed data show that for the trabecular region of interest (ROI) chosen, the mice showed no signi cant differences between the CON and the OB groups in any of the trabecular bone parameters measured. By contrast with mice, the PRE group of ground squirrels, as compared with the SA group, showed both greater bone surface density (BS/BV) and trabecular number (Tb.N) values that were signi cantly higher in the PRE group (by 36.4% and 28.6%, respectively, P < 0.05) as compared with SA animals. However, the trabecular thickness (Tb.Th) and trabecular separation (Tb.Sp) values were both reduced signi cantly in the PRE group (by -31.6% and − 24.3%, respectively, P < 0.05) as compared with SA squirrels. Finally, the bone volume fraction (BV/TV, the ratio of bone tissue volume to tissue volume) showed no signi cant differences between the SA and the PRE ground squirrels. Values are mean ± SD, n = 6. *P < 0.05 compared with CON group. # P < 0.05 compared with SA group.
Cortical bone quantitative parameters are also presented in Table 4. In mice, there were no signi cant differences in any of measured parameters between the CON and the OB groups. Among ground squirrels, as compared with the SA group, the PRE group showed signi cant reductions in average cortical thickness (Ct.Th), marrow area (Ma.Ar), cortical bone area (Ct.Ar) and total area (Tt.Ar) by -52.2%, -88.6%, -88.7% and − 88.6%, respectively (P < 0.05). Table 4 also shows an analysis of bone mineral-related data. Compared with the CON group, the tissue mineral density (TMD) and tissue mineral content (TMC) showed rising trends in the OB group (P = 0.095 and 0.070, respectively). The bone mineral density (BMD) of the OB group did not change signi cantly as compared with the CON mice. And bone mineral content (BMC) in the OB mice increased signi cantly by 1.2-fold. Ground squirrels showed a very different pro le. No signi cant difference in tissue mineral density (TMD) was found between the PRE and SA ground squirrels. However, compared with the SA group, the tissue mineral content (TMC) in the PRE group bone was signi cantly and strongly reduced by 82.8% (P < 0.05). There was no signi cant difference in bone mineral density (BMD) between the PRE and SA ground squirrels but bone mineral content (BMC) of the PRE group was signi cantly reduced by 81.9% compared with the SA group (P < 0.05). Figure 1 also shows representative micro-CT images of the proximal tibia in the different groups and Table 5 shows quanti ed parameters for tibial bone. The mouse images (Fig. 1) show no visual differences between the CON and the OB groups. This is backed up by the lack of signi cant change by any of the ve parameters assessed (Table 5). By contrast, in the ground squirrels, of the parameters measured for trabecular bone only bone volume fraction (BV/TV) increased signi cantly in the PRE group, by 40.4% (P < 0.05). However, the Tb.Sp was reduced signi cantly in the PRE group of ground squirrels (-33.0%, P < 0.05) whereas the Tb.N value showed an increasing trend but did not reach signi cance in the PRE group compared with the SA group (P = 0.064). Mineral-related data for tibial bone were also analyzed and the results are shown in Table 5. In the OB mice, there were no signi cant differences in any of the four tibial bone mineral parameters as compared with the CON mice. There was no signi cant difference in tissue mineral density (TMD) or tissue mineral content (TMC) of the PRE group as compared with the SA group in ground squirrel as well as no change in bone mineral content (BMC) between the PRE and SA group. However, the bone mineral density (BMD) of the PRE group was signi cantly higher by 20.1% as compared with the SA group of ground squirrels (P < 0.05).

Three-point bending test
The results of the three-point bending test of the femurs are shown in Table 6A. In mice, neither the ultimate bearing capacity (N) nor the ultimate bending energy (J) differed between CON and OB mice but the stiffness parameter (N/mm) was signi cantly greater by 32.7% in the OB mice group (P < 0.05). Comparable analysis of bone from SA versus PRE ground squirrels found no signi cant differences in ultimate bearing capacity, stiffness or ultimate bending energy between the two ground squirrel groups.
The results of the three-point bending test for tibia are shown in Table 6B. The ultimate bearing capacity, stiffness and the ultimate bending energy were not signi cantly different between CON and OB groups of mice. The same was true of the comparison between SA and PRE ground squirrels although increasing trends were noted in the PRE group for the ultimate bearing capacity (P = 0.051) and stiffness (P = 0.095).
Similarly, there was no signi cant difference in ultimate bearing capacity, stiffness and ultimate bending energy between the OB and the CON mice. Values are mean ± SD, n = 6. *P < 0.05 compared with CON group.

Relative protein expression levels
We used Western blotting to detect the expression levels of proteins related to bone formation, and the results are shown in Fig. 2. Compared with the CON mice, the expression level of RunX2 was not signi cantly different in the OB group ( Fig. 2A, B). However, the expression levels of OCN and ALP showed small but signi cant increases between the OB and the CON mice (20.2% and 21.9%, respectively, P < 0.05) ( Fig. 2A, C, D). By contrast, the expression level of RunX2 in PRE ground squirrels was signi cantly higher than in the SA group (27.2%, P < 0.05) ( Fig. 2A, B), whereas the expression of OCN decreased in the PRE ground squirrel group (-38.4%, P < 0.05) ( Fig. 2A, C). ALP expression level increased signi cantly between the SA and the PRE ground squirrels (26.0%, P < 0.05) ( Fig. 2A, D).
The expression levels of proteins related to bone resorption are shown in Fig. 3 Proteins associated with Wnt signaling were also assessed by Western blots (Fig. 4). Compared with the CON group, the expression levels of P-β-catenin and GSK-3β in the OB group were signi cantly increased by 20.2% and 38.8%, respectively (P < 0.05). The expression level of GSK-3β was also signi cantly increased in the PRE group compared with the SA group (2.4-fold, P < 0.05), but P-β-catenin did not change between SA and PRE ground squirrels.

Discussion
In this study, we innovatively compared the differences in bone metabolism between pathological obesity (in mice) induced by a high-fat diet and healthy obesity (in ground squirrels) induced by natural fattening before hibernation. We measured body weight, adipose tissue wet weight, bone microstructure, bone mechanical properties, and protein expression levels related to bone formation, bone resorption and Wnt signaling. The results showed that obese mice (high-fat diet) showed no obvious alterations or abnormalities in bone microstructure as compared with controls, but bone strength increased and the expression levels of proteins related to bone formation and bone loss increased and maintained a dynamic balance. By contrast, bone formation was enhanced in pre-hibernation ground squirrels, which was manifested as an enhancement of bone microstructure, improvement of bone strength, increased expression levels of bone formation-related proteins (RunX2 and ALP), increased expression levels of bone resorption-related proteins, and enhanced Wnt signaling.
The body weight and adipose tissue wet weight of OB mice and PRE ground squirrels were signi cantly increased compared with CON mice and SA ground squirrels, respectively ( Table 2 and Table 3). After being fed a high-fat diet, the body weight of mice in the OB group increased signi cantly by 10.6% compared with the CON group. However, after natural fattening, the body weight of ground squirrels in the PRE group increased much more with a signi cant 62.5% increase over the SA group (Table 2). This indicated that the degree of obesity in pre-hibernating ground squirrels was much greater than in mice. At the same time, our data showed that after fattening, the change in adipose tissue of mice occurred in perirenal and mesenteric adipose, whereas in ground squirrels, adipose accumulation was mainly in mesenteric and subcutaneous depots with lesser accumulation of perirenal adipose. There is so little subcutaneous adipose in mice that it cannot be easily separated by surgical procedures. In fact, we can't deny that subcutaneous adipose is always present and may be extracted by other methods. Previous studies have shown an inverse relationship between visceral fat and bone density [52]. Therefore, compared with mice, the characteristics of less visceral fat exhibited by ground squirrels may be related to the increase in bone density.
There was no signi cant change in the microstructure of the femur or tibia in the OB group, but the bone mineral parameters BMC of mice showed a signi cant increase, TMD and TMC showed an increasing trend (P = 0.095 and 0.070, respectively) in femur after fattening (Table 4). This indicated that the bone microstructure of mice was in a balanced state, but bone minerals had a tendency to increase in femur. By contrast, ground squirrels showed some substantial differences between SA and PRE states. The BS/BV and Tb.N parameters of the PRE group were signi cantly increased by 36.4% and 28.6%, respectively, whereas Tb.Th, Tb.Sp, Ma.Ar, TMC and BMC were all strongly reduced by 31.6%, 24.3%, 88.6%, 82.8% and 81.9%, respectively, indicating that the bone formation of the femur in the PRE group was enhanced ( Table 4). The microstructural changes of the ground squirrel tibia was similar to those of the femur. The Tb.N showed an increasing trend (P = 0.064), Tb.Sp was signi cantly decreased (-33.0%, P < 0.05), and BV/TV and BMD were signi cantly increased (40.4% and 20.1%, respectively, P < 0.05), which indicated that the bone formation was enhanced in the tibia (Table 5). In conclusion, there was no tissue speci city in bone formation on the femur and tibia between the mice and ground squirrels. This is consistent with another study, which shows that obese Wistar rats induced by high-fat diet have no difference in bone formation between femur and tibia [53]. Compared with the mice, the bone mass in the PRE group increased in both femur and tibia of ground squirrels, which showed that healthy obesity was not harmful to their bones. However, interestingly, the bone minerals of the femurs in the two models showed opposite changes, an increase in the OB mice and a decrease in the PRE ground squirrels. The mineral density is related to the mechanical properties of bones and therefore, we postulated that these two types of obesity have opposite effects on the mechanical properties of bones. Hence, we next determined the mechanical properties of bones from the two obesity models.
For this study, we recorded changes in the mechanical properties of the bones of mice fed the high-fat diet for 3 months and of ground squirrels fattened before hibernation. Using a three-point bending test, the stiffness of the femur in the OB group was signi cantly increased by 32.7%, but the mechanical properties of the tibia did not change signi cantly (Table 6). This indicated that the increased bending resistance of the OB mice was mainly manifested in the femur, thereby reducing the risk of fracture. This may be an adaptation to the higher load caused by weight gain. Compared with the SA ground squirrels, the bones of the PRE group also showed different mechanical properties. The ultimate bearing capacity, stiffness and ultimate bending energy of the femur in the PRE group did not change signi cantly compared with SA group (Table 6), which indicated that the mechanical properties of the femur did not change during the approach to the hibernation season. However, the ultimate bearing capacity and stiffness of the tibia in the PRE group showed an increasing trend (P = 0.051 and 0.095, respectively, Table 6), which suggests that an increase in bending resistance in the PRE group was mainly manifested in the tibia, which could also reduce the risk of fracture. The changes of bone microstructure and mechanical characteristics were related to changes in bone remodeling function. Therefore, we examined the expression levels of key proteins that regulate bone formation, bone resorption, and Wnt signaling pathways.
Bone metabolism is maintained by the dynamic balance of bone formation and bone resorption [19]. RunX2 is the main driving factor of bone formation and promotes the differentiation and maturation of osteoblasts [54]. The expression level of RunX2 in the OB mice did not change signi cantly as compared to CON group (Fig. 2B), which was different a previous study that showed a signi cant decrease in RunX2 mRNA levels in 4-week-old male rats fed with a high-fat diet for 22 weeks [55]. By contrast, the expression level of RunX2 in PRE ground squirrels was signi cantly up-regulated compared with SA group, indicating that bone formation in the PRE group was enhanced. The differential expression of RunX2 may be the reason for the different changes in bone microstructure between the PRE ground squirrels and the OB mice. OCN plays an important role in regulating calcium metabolism of the bone, mainly promoting bone mineralization [56]. In this study, the expression level of OCN in OB mice was up-regulated, whereas the expression level of OCN in PRE ground squirrels was down-regulated (Fig. 2C). This indicates that the OB mice had increased bone mineralization ability, whereas the PRE ground squirrels could be decreasing bone mineralization activity as the hibernation season approaches. In addition, OCN not only plays a role in bone formation, but also affects energy regulation [57] and, hence, the different changes in the expression of OCN in OB mice and PRE ground squirrels may also contribute to differential regulation of energy metabolism. This idea requires further experimental. ALP protein is one of the phenotypic markers of osteoblasts and can directly re ect the activity or function of osteoblasts [58]. The expression level of ALP in both the OB mice and the PRE ground squirrels was signi cantly increased (Fig. 2D), which indicated that the osteoblasts in both groups were in good activity and function. This is consistent with a previous study on 6-week-old male C57BL/6 mice fed a high-fat diet for 14 weeks, the results showing that the expression level of ALP in obese mice was signi cantly increased [59]. Studies have shown that ALP can promote the absorption of calcium ions by bones [60]. We speculate that due to insu cient obesity in the high-fat OB model in the present study, only a 10.6% weight gain was achieved compared with the controls. In order to adapt to the higher load caused by moderate obesity, the bones signi cantly increased the expression level of ALP, thereby promoting the absorption of calcium salts by the bones, increasing the bone minerals, and enhancing the mechanical strength of the bones. This was also in line with the increasing trend of TMD, TMC and BMC obtained by Micro CT in this study. Hence, we propose that the difference in the expression level of RunX2 was a main reason for the difference in bone formation between the two models. Compared with the OB group, the bone formation in the PRE group was at a higher level.
In terms of bone resorption, the expression levels of RANKL, Cathepsin K and MMP9 increased signi cantly in both the OB mice and the PRE ground squirrels (Fig. 3). The enhancement of bone resorption in obese mice was consistent with previous studies [34,61,62]. We speculated that the reason why bone loss did not occur in the OB mice was that both bone formation and bone resorption were upregulated to achieve a dynamic balance of high expression. Although there was no bone loss in the OB mice, the high expression levels of bone resorption proteins may be a potential risk for bone loss in mice.
Studies have shown that bone loss occurred in mice when they were extremely obese [63]. The expression levels of bone resorption proteins were signi cantly increased in the PRE group of ground squirrels, but the bone substance was also increased, which may be caused by greater bone formation than bone resorption.
In addition, Wnt signaling also plays an important role in the regulation of bone remodeling [64]. In the present study, the expression levels of P-β-catenin and GSK-3β in OB mice were signi cantly increased (38.8% and 20.2%, respectively, Fig. 4), and Wnt signaling was weakened, which could lead to an increase of bone resorption and a decrease of bone formation, which is consistent with a study that showed that obesity inhibited the Wnt signaling pathway [65]. Different responses were seen in ground squirrels, where the expression level of GSK-3β in the PRE group was signi cantly increased (1.4-fold, Fig. 4C), but the expression level of P-β-catenin did not change (Fig. 4B), which indicated that the Wnt signal was strengthened and bone formation was enhanced before hibernation [47]. The differential expression of Wnt signals in the two models was also the cause of bone changes. Activation of typical Wnt signaling by inhibiting GSK-3β has been shown to increase bone mass, which may involve many mechanisms[66].
However, although GSK-3β inhibitors can promote osteogenesis, we should note that the activity of GSK-3β is not only manifested in osteogenesis, but is also related to other intracellular biological processes, which has raised concerns about possible side effects of long-term treatment with these inhibitors in humans[67]. In addition, over-inhibition of GSK-3β has the risk of tumorigenicity [68].
When comparing the two models, we found that weight gain will cause a signi cant increase in the expression of bone resorption proteins in both the OB mice and the PRE ground squirrels. The bone substance of the mice did not change signi cantly, which may be caused by an unchanged expression level of RunX2 and the signi cant increases in the expression levels of OCN and ALP. Although body weight of the OB mice only increased by 10.6% compared with the control group, the weight gain also brought a great risk of bone loss to the OB mice, which was manifested as a signi cant up-regulation of bone resorption proteins and weakened Wnt signaling. Different from mice, ground squirrels showed different regulatory mechanisms at work. Although the expression levels of bone resorption proteins in the PRE group also increased signi cantly, the protein expression levels of RunX2, ALP and GSK-3β increased signi cantly, resulting in greater bone formation than bone resorption and a net increase in bone mass. This mechanism, which is different from pathological obesity, suggests that ground squirrels fattened before hibernation can be studied as an anti-obesity bone loss model.

Conclusions
In summary, we compared the differences in bone metabolism between high-fat diet fattened mice and naturally fattened ground squirrels. Our study shows that the hind limb bones of mice in the OB group did not undergo bone loss, and the femurs of the ground squirrels in the PRE group developed bone formation. And the results of the three-point bending test showed that the skeletal mechanical properties of the OB mice were strengthened, while the PRE ground squirrels did not change signi cantly. Western Blot showed that the levels of proteins related to both bone formation and bone loss were up-regulated in the OB mice, and bone metabolism was at a higher level of metabolic balance. For pre-hibernating ground squirrels, the healthy obesity acquired before hibernation also enhanced the expression levels of proteins related to bone formation, bone resorption and Wnt signaling.

Declarations
Ethics approval and consent to participate All animal experiments were approved by the Experimental Animal Protection Committee of the Ministry of Health of the People's Republic of China.

Consent for publication
Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  pre-hibernation, squirrels that nished natural fattening, sacri ced in late-autumn (end of September, 30-40 d before hibernation). Values are mean ± SD, n = 6. *P < 0.05 compared with CON group. #P < 0.05 compared with SA group.