High Fat Diet-Induced Obesity Negatively Affects Whole Bone Bending Strength but not Cortical Structure in the Femur

Although body mass index is positively associated with bone mineral density, suggesting obesity is protective against fracture, elderly obese individuals experience greater fracture risk at certain sites than non-obese peers, suggesting bone structural or material changes contribute to fragility. Diet-induced obesity rodent studies have reported detrimental changes to bone microstructure and some apparent-level material properties, but tissue-level material changes are not well understood. Because adipose tissue is highly vascularized, and bone remodeling depends critically on functional vascular supply, concurrent effects on osteovascular perfusion and structure may provide insight about obesity-related bone fragility. This study aimed to determine the effects of obesity on both tissue-level bone properties and osteovascular properties that could negatively impact bone strength. Five-week-old male C57Bl/6J mice were fed either high fat diet (HFD) or control fat diet (CFD) for 17 weeks and received daily treadmill exercise or remained sedentary for eight weeks at ages 14-22 weeks. HFD negatively affected femur bending strength, with 18% lower yield load than CFD. Although HFD negatively altered cancellous microstructure in the distal femur, with 32% lower bone volume fraction than CFD, it did not affect cortical bone geometry in the femoral metaphysis or diaphysis. HFD caused increased carbonate substitution but had no effect on other composition metrics or apparent- or tissue-level material properties in the femoral diaphysis. Exercise did not affect bone strength or microstructure but increased endosteal mineralizing surface in the tibial diaphysis, mineral crystallinity and mineral-to-matrix ratio in the femur, and blood supply to the proximal tibial metaphysis. HFD did not affect blood supply in the tibia or 2D osteovascular structure in the distal femoral metaphysis, indicating that HFD negatively affects cancellous bone without affecting osteovasculature. This study reveals that HFD negatively affected cancellous microstructure without affecting osteovascular structure, and whole-bone strength without altering cortical geometry or material properties.

HFD negatively affected cancellous microstructure without affecting osteovascular structure, and

Introduction
The right femur underwent three-point bending to failure to measure whole bone mechanical 164 properties and estimated apparent-level material properties. Immediately prior to testing, the femur 165 was brought to room temperature and placed in a 37°C bath of 1X PBS for 60 sec. The bone was 166 centered over a 6.5-mm lower span (40% average femur length) with the anterior side facing up 167 so that the anterior diaphysis was loaded in compression. Three-point bending was performed to  (40) Yield was calculated as the point where a line with a 5% decrease in stiffness intersected 175 the force-displacement curve. (40) PYD was calculated as the difference between the deformation at 176 yield and the deformation at failure. The stress-strain curve was estimated using the cross-sectional  Vascular structure and proximity to bone surfaces were examined in the distal femoral metaphysis 231 using thick-section immunofluorescence to quantify the amount of blood vessels, labeled by 232 endomucin (EMCN), and bone surfaces, labeled by collagen type I (COL-1). (47) The remaining distal portion of the right femur samples were fixed overnight in 10% neutral buffered formalin at     For analysis #2, outcome parameters were compared between diet and activity, with interaction, 287 using two-way analysis of variance (R 'aov' function). Tukey's post-hoc tests were used to 288 compare group means. Vascular structure parameters were analyzed with a similar model, but the 289 interaction between diet and activity was not modeled due to missing data and thus insufficient 290 power to analyze the full model. Three-point bending parameters were further analyzed with two 291 analysis of covariance (ANCOVA) models, one with mass as the continuous covariate and one 292 with femur length as the continuous covariate. (40,48) 293 294 For analysis #3, the same repeated measures factorial model used in analysis #1 was used (SAS 295 'MIXED' procedure), but parameters were compared between diet and activity groups across scan 296 region (anterior and posterior for nanoindentation; posterior, lateral, anterior, and medial for 297 Raman spectroscopy). The residual variance was modeled assuming compound symmetry 298 covariance. Predicted least-squares means with Tukey-Kramer adjustment for multiple 299 comparisons were used to analyze pairwise differences between diet and activity groups, with 300 interaction (i.e., HFD-Sedentary vs. HFD-Exercise).

303
Weekly measures of mouse mass, serum glucose, and monthly glucose tolerance tests confirmed 304 that the high fat diet produced an obese phenotype in this study. The HFD group had consistently 305 greater body mass at all timepoints compared to the CFD group (p = 0.0016, Figure 3A). At the 306 end of the study, after 17 weeks of diet, the HFD group (43.0 ± 5.2 g) weighed 33% more than the 307 CFD group (32.4 ± 1.7 g, p < 0.0001). Overall, the HFD group had increased fasting glucose 308 concentrations relative to the CFD group (p = 0.0054), but not at every timepoint ( Figure 3B). At  At the end of the study, in vivo perfusion in the proximal tibial metaphysis was 29% greater in 331 exercise groups (12.2 ± 3.0 PU) compared to sedentary groups (9.5 ± 1.6 PU, p = 0.044, Figure 4).

332
Tibial perfusion was similar between HFD (11.7 ± 2.6 PU) and CFD groups (10.1 ± 2.7 PU, p = 333 0.23). ). In addition, the connectivity density (Conn.D) of the trabecular network was 50% lower in 343 the HFD group compared to the CFD group (p = 0.00053, Table 1), but the degree of anisotropy 344 (DA) was not significantly different between HFD and CFD groups (p = 0.11, Table 1). HFD  (Table 1). Similarly, in the mid-diaphyseal VOI, neither HFD nor exercise affected  (Table 2). Exercise, however, did significantly affect the extent of  (Table 2).  Table 3). Compared to the CFD group, the HFD group had 18% lower yield load ( p = 0.039) and nearly lower ultimate load (14% lower, p = 0.058) and stiffness (18% lower, 371 p = 0.055). After accounting for body mass (ANCOVA), none of the mechanical properties 372 differed between HFD and CFD groups, except whole bone stiffness tended to be lower in HFD 373 compared to CFD even after body mass adjustments (p = 0.085, Table 3). Femoral length was 374 similar across diet and exercise groups (Table 1), but when whole bone mechanical properties were 375 adjusted for femur length (ANCOVA), none of the mechanical properties differed between HFD 376 and CFD groups, except yield load was nearly lower in HFD compared to CFD (p = 0.082, Table   377 3). Similarly, none of the estimated apparent-level material properties -yield stress, ultimate 378 stress, and Young's modulus -were significantly affected by HFD or exercise (Table 3). Cortical 379 tissue material properties assessed with nanoindentation were also not significantly affected by 380 HFD or exercise (Table 3)   Mineral-to-matrix band intensity ratios near the endosteal edge were higher for exercise groups 391 relative to sedentary groups for the phosphate/(proline+hydroxyproline) ratio (27% higher, p = 392 0.013, Figure 7B), phosphate/amide I ratio (18% higher, p = 0.030, Figure 7C), and phosphate/amide III ratio (25% higher, p = 0.023, Figure 7D). Similarly, the carbonate-to-matrix 394 ratio (carbonate/amide I) near the endosteal edge was also increased for exercise compared to 395 sedentary (13% higher, p = 0.023, Figure 7E). Carbonate substitution was not affected by exercise 396 in any region ( Figure 7F).

417
High fat diet-induced obesity reduced whole bone bending strength in the femur, without altering 418 cortical bone mineral density, geometry, or apparent-or tissue-level material properties relative to 419 control fat diet. Because bone strength depends on these parameters, (9,10) we expected one of them 420 to be altered by HFD to explain the underlying cause for the relative strength deficits in that group.

421
The reductions in bending properties with HFD were no longer significant after adjusting for body HFD created an obese, hyperglycemic phenotype that persisted with daily treadmill exercise. After 439 9 weeks of diet, HFD groups were heavier than CFD groups, and after 13 weeks of diet, HFD 440 groups had significantly lower glucose tolerance and weekly fasting serum glucose concentrations 441 that were over 200 mg/dL, indicative of pre-diabetes. (51) Although exercise had transient benefits 442 to glucose tolerance in the HFD group, these benefits did not persist to the end of the study at 443 Week 17, and daily treadmill exercise did not mitigate the negative effects of HFD on cancellous 444 bone microstructure. Exercise had no effect on femoral cortical mechanical properties at the whole 445 bone, apparent, or tissue levels, despite slightly increasing mineral-to-matrix ratios in the 446 diaphysis. Exercise, but not HFD, increased the extent of active remodeling bone surface in the 447 tibial diaphysis and bone perfusion in the proximal tibia but had no effect on the relative amount 448 of blood vessels or the distance between blood vessels and bone surfaces in the distal femur. These studies demonstrate that diet-induced obesity in male mice commonly leads to detrimental 468 changes in cancellous bone microstructure, as we report here, and suggest that altered modeling 469 during skeletal growth is not solely responsible for the negative HFD effects on microstructure.

471
The effects of HFD on cortical bone geometry in male C57Bl/6J mice are less consistent. Similar 472 to our results, several groups report no effect on cortical bone parameters, (13,14,16,54) but the study 473 that reported increased trabecular cross-sectional area also found a 7% increase in cortical area in  Studies with HFD beginning at 3-6 weeks of age and ending at 19-28 weeks of age reported a 12% 493 reduction in maximum load, (19) 29% reduction in ultimate load, and 20% reduction in stiffness. (52)

494
Similar results have also been reported in the L3 vertebra, with mice fed a 60% HFD from 5-17 495 weeks of age (young) or from 20-32 weeks of age (mature) having 17-24% lower yield load, 16-496 26% lower maximum load, and 21-27% lower stiffness during compressive loading in both age 497 groups compared to age-matched mice fed a CFD. (15) Conversely, in a study of cantilever bending 498 in the femoral neck, the HFD group (60% fat diet from 7-28 weeks of age) had 18% higher 499 maximum load and 29% higher bending modulus compared to the CFD group.  For whole femurs in three-point bending, two studies found that HFD (60% fat diet starting from 3-6 weeks-of-age to 19-28 weeks of age) caused 19-32% lower apparent elastic modulus, 15-26% 508 lower maximum stress, and 24% lower yield stress, (18,19) while another study found 44% higher 509 apparent elastic modulus. (52) Tissue-level material properties were also unaltered by HFD in our 510 study. To our knowledge, no previous study has examined the effects of HFD on tissue-level 511 material properties. Since we did not find HFD-induced changes in bone density, structure, or 512 tissue-level properties, the reduced whole bone strength may result from a combination of small 513 changes in several parameters that were not statistically significant in this study.

515
Cortical tissue composition in the femur was altered by exercise, with increased mineral-to-matrix 516 and carbonate-to-matrix ratios near the endosteal edge and increased mineral maturity near the 517 periosteal edge. Mineralization of new bone tissue occurs slowly, so higher mineral-to-matrix and 518 carbonate-to-matrix ratios are associated with older bone that is generally harder and stiffer. (45,60) 519 However, a study using the same treadmill regimen initiated at 16 weeks of age found that 520 treadmill exercise increased ultimate strain and the mineral-to-matrix ratio of phosphate v1 to 521 summed proline and hydroxyproline without affecting tibial morphology, suggesting increased 522 mineral-to-matrix ratios could be a mechanism by which bone adapts to exercise to maintain local However, the Raman spectra in our study did not contain any of these AGE bands, indicating 532 AGEs were not significantly present.

534
This study found no effect of HFD or exercise on 2D osteovascular structure (vessel area and 535 proximity to bone surfaces) in the distal femur, but stereological methods are not ideal for 536 measuring complex three-dimensional structures like the branching network of blood vessels, (67) 537 so HFD may have affected osteovascular parameters that are not quantifiable with stereology. Diet-induced obesity has far-reaching physiological effects that can impact bone health and may 557 be responsible for the observed envelope-specific changes to cancellous but not cortical bone 558 structure. In this study, HFD led to the development of obesity and pre-diabetic levels of elevated 559 serum glucose, both of which impact metabolic pathways that influence bone metabolism. obesity on increased serum glucocorticoids in either rodents or humans is unclear. (75,79) Lastly, 575 increased amounts of marrow adipose tissue (MAT) may negatively affect cancellous bone structure. We did not quantify MAT in this study, but other groups report dramatic increases in the 577 amount of metaphyseal MAT with HFD, (13,(24)(25)(26)(27) and decreased MAT with intense exercise. (24,25) 578 Moderate treadmill exercise did not affect bone microstructure in this study, but other studies that 579 utilize more intense exercise regimen, such as free access to running wheel (24,25,53,80) or high 580 intensity treadmill training, (81) found effects of exercise in HFD mice.

582
In conclusion, our study demonstrated that high fat diet-induced obesity caused detriments to 583 cancellous bone microstructure and whole bone bending strength in the femur that were not 584 concomitant with changes to metaphyseal perfusion or vascularity, or to cortical geometry or tissue 585 properties. We also showed that moderate treadmill activity did not reverse the deleterious effects 586 of HFD, increase intraosseous vascularity, or increase mechanical properties in this model.

587
Exercise did, however, increase intraosseous perfusion in the tibia, and stimulate changes to tissue        proline and hydroxyproline ratio, C) phosphate v1 to amide I ratio, D) phosphate v1 to amide III ratio, and E) carbonate v1 to amide I ratio but not F) carbonate substitution. Points represent mean of all quadrants per femur, lines and bars represent estimated least-squares mean ± 95% confidence interval c: p < 0.05 Ex vs. Sed (main effect), d: p < 0.10 HFD vs CFD (main effect), g: p < 0.10 Ex vs. Sed (main effect).