Circadian rhythm disruption with high-fat diet impairs glycemic control and bone quality

Abstract

Biological functions, including glycemic control and bone metabolism, are highly influenced by the body’s internal clock. Circadian rhythms are biological rhythms that run with a period close to 24 hours and receive input from environmental stimuli, such as the light/dark cycle. We investigated the effects of circadian rhythm disruption (CRD), through alteration of the light/dark schedule, on glycemic control and bone quality of mice. Ten-week-old male mice (C57/BL6, n=48) were given a low-fat diet (LFD) or a high-fat diet (HFD) and kept on a dayshift or altered schedule (RSS3) for 22-weeks. Mice were divided into four experimental groups (n=12/group): Dayshift/LFD, Dayshift/HFD, RSS3/LFD, and RSS3/HFD. CRD in growing mice fed a HFD resulted in a diabetic state, with a 36.2% increase in fasting glucose levels compared to Dayshift/LFD group. MicroCT scans of femora revealed a reduction in inner and outer surface expansion for mice on a HFD and altered light schedule. Cancellous bone demonstrated deterioration of bone quality as trabecular number and thickness decreased while trabecular separation increased. While HFD increased cortical BMD, its combination with CRD reduced this phenomenon. The growth of mineral crystals, determined by SAXS, showed HFD led to smaller crystals. Considering modifications of the organic matrix, regardless of diet, CRD exacerbated the accumulation of fluorescent advanced glycation end-products (fAGEs) in collagen. Strength testing of tibiae showed that CRD mitigated the higher strength in HFD group and increased brittleness indicated by lower post-yield deflection and work-to-fracture. Consistent with accumulation of fAGEs, various measures of toughness were lowered with CRD, but combination of CRD with HFD protected against this decrease. Differences between strength and toughness results represent different contributions of structural and material properties of bone to energy dissipation. Collectively, these results demonstrate that combination of CRD with HFD impairs glycemic control and have complex effects on bone quality.

Introduction

Most biological processes in mammals are highly influenced by the body’s internal clock.^(1,2) The circadian system consists of an input to the master clock that is translated into specific rhythms for maintaining homeostasis.⁽²⁾ Circadian rhythms are near 24 hour biological rhythms.^(2,3) The master clock generates and regulates all circadian rhythms and the light/dark cycle is the strongest synchronizer of circadian rhythms to the 24-h day.⁽³⁾ The master clock provides timing signals to the peripheral clocks is found in all tissues, including bone.^(4–8) The molecular clocks of bone tissue play an important role in the proper maintenance of the bone turnover rate,⁽⁴⁾ because they influence the core clock genes found in bone-forming osteoblasts (OBs) and bone-resorbing osteoclasts (OCs).^(5–7) These core clock genes also control steroids, such as endogenous glucocorticoids, which are involved in the regulation of lipid metabolism, glucose homeostasis and play an important role in overall bone metabolism.⁽⁸⁾

It has been shown that the proper timing of these circadian rhythms is needed to ensure metabolic control.^(1,2) Pre-clinical and clinical studies have shown that circadian rhythm disruption (CRD) is linked to loss of metabolic control and may lead to hyperglycemia ^(9–12) and skeletal fragility.^(4–8) Clinical studies have also shown that CRD is a contributing factor in the development and progression of different diseases, for example, Type 2 Diabetes (T2D)⁽¹¹⁾ and Alzheimer’s Disease.^(13–14) CRD contributes to insulin resistance and B-cell dysfunction, increasing the risk of T2D.^(9–10) In particular, studies using mouse models demonstrated that CRD leads to hyperglycemia because of an increase in glucose levels⁽⁹⁾ similarly to high-fat diets (HFDs).^(15–16) The excess of blood glucose has been demonstrated to increase the formation of fluorescent advance glycation end-products (fAGEs) which are produced by non-enzymatic glycation (NEG), also known as the Milliard reaction.⁽¹⁷⁾ NEG typically occurs between reducing sugars such as glucose and proteins, DNA, and lipids^(18,19). Increase in the formation of fAGEs has been strongly linked to the loss of bone heath and increased skeletal fragility. ^(17–27)

Considering clinical relevance, disrupted circadian rhythm and diets rich in fat and sugar continue to be prevalent amongst children.⁽²⁸⁾ However, the combination of circadian rhythm disruption and a HFD during growth with a focus on bone quality has not been established. Therefore, the goal of this study was to evaluate the effects of both CRD and HFD on bone quality during growth using a mouse model. To this end, we compared glycemic control and bone quality in young mice fed either a high-fat or low-fat diet and kept on a regular or disrupted schedule until adulthood. We hypothesized that combination of HFD with CRDs would impair glycemic control and degrade bone quality.

Materials and Methods

Experimental design

Forty-eight male mice (C57/BL6) were obtained from Taconic Farms and transferred to the BioResearch Center at Rensselaer Polytechnic Institute at 8 weeks old. Animals were housed in two rooms, twenty-four mice per room with four mice per cage. Prior to arrival at our animal facility, all animals were exposed to 12-hour light:12-hour dark (12L:12D) lighting condition. After arriving to our facility, all animals continued to be exposed to 12L:12D for 2 consecutive weeks to acclimate them to the cages. For the remaining 22 weeks of the study, one room was kept on a 12L:12D (Dayshift) schedule while the other group was exposed to 3 consecutive simulated nights of rotating shift work per week (RSS3) with a light schedule reversal as follows: Monday (12L:12D), Tuesday (12L:12D), Wednesday (12L:12D), Thursday (12L:12D), Friday (12D:12L), Saturday (12D:12L), Sunday (12D:12L). One group per room was administered a high-fat diet (HFD; 46% fat, 36% carbs, 18% proteins), while the other one was given a low-fat diet (LFD; 10% fat, 72% carbs, 18% proteins). Food was purchased from TestDiets (St. Louis, MO) and can be found under catalog #58Y2 (LFD), and #58V8 (HFD). The cage-lighting system was developed and established by Figueiro et al.⁽²⁹⁾ Each light fixture contained the same green LEDs (peak wavelength = 519 nm, full-width half-maximum bandwidth = 40 nm). All procedures were approved and carried out according to the rules and regulations of the Rensselaer Polytechnic Institutes’ Institutional Animal Care and Use Committee.

Body mass and glucose measurements

Body mass measurements, fasting blood glucose levels and oral glucose tolerance tests (OGTT) were performed every 7 weeks, on average, from the start of the treatment until 32 weeks of age. The mice fasted for 6 hours prior to the start of the OGTT, and each test was done during the light period. Prior to the first glucose dose, the fasting blood glucose levels were measured. The glucose dose for the OGTT test was based on the body mass of each animal; 2 grams glucose per kilogram of body mass was administered through a feeding needle and blood was collected during a period of two hours at the 0, 15-, 30-, 60-, and 120-minute time points⁽³⁰⁾. Fasting (baseline) blood glucose measurements were used to determine whether the animals were in the normal (blood glucose < 199 mg/dL), pre-diabetic (blood glucose: 200–249 mg/dL), or the diabetic range (blood glucose > 250 mg/dL).⁽³¹⁾ The time of feeding, body mass and OGTT measurements were varied to prevent mice from adapting to these cues and limit impact on circadian entrainment.⁽²⁹⁾

Tissue collection and harvest

Animals were euthanized at 32 weeks of age by isoflurane overdose. Cervical dislocation was used as a secondary method of confirmation. Three animals that died during OGTT tests were excluded from the study. All skin and soft tissues were removed, and femora and tibiae were harvested for analysis. Samples were kept in Eppendorf tubes, submerged in saline, and stored at −80^oC.

Micro-computed tomography (μCT) analysis

All femora and tibiae were scanned using a μCT scanner (VivaCT40, Scanco Medical AG, Basserdoft, Switzerland). High resolution scans were acquired at 70 kVp energy, 114 mA current, 301 ms integration time using a 10.5 mm voxel size. Custom evaluation scripts with pre-defined volume of interest (VOI) were used to analyze the cortical and trabecular morphology of femora and tibiae, respectively. For the trabecular analysis of tibiae, the VOI was defined by first locating the growth plate in the distal end of the tibiae. Contouring was started 5 slices from it; with new contours created every 10 slices for a total of 100 slices (approximately 1.05–1.25 mm) moving in the proximal direction from the growth plate and evaluating the secondary spongiosa region. Contoured slices were then morphed and reduced by 5% to ensure that cortical bone was not included in the analyses. This analysis yielded the trabecular thickness (Tb. Th), trabecular number (Tb. N), trabecular separation (Tb. Sp), and bone volume fraction (BV/TV). For the cortical morphology of femora, the VOI was defined from an approximately 0.3 mm region of the femora in the midshaft. This analysis yielded the endosteal diameter, periosteal diameter, cortical thickness (Ct. Th), moment of inertia (I_xx), and bone mineral density (BMD). To determine the cortical morphology of the tibiae, the whole-bone was scanned to find the mid-point and a VOI of an approximately 0.5mm region from the midshaft was evaluated. This analysis yielded the endosteal diameter, periosteal diameter, cortical thickness (Ct. Th), moment of inertia (I_xx), and bone mineral density (BMD).⁽³²⁾ Bone mineral density was calculated with a hydroxyapatite (HA) phantom used for calibration,⁽³²⁾ and global threshold for all scans (femora and tibiae) was set for 656 mg HA/cm³ with a gaussian filtration (σ = 0.8, support = 1).

Small Angle X-Ray Scattering (SAXS)

Three representative femora samples per experimental group were longitudinally sectioned from the mid-diaphyseal region to a uniform thickness (300 μm) and then set in epoxy (AstroChem 1119). Two-dimensional SAXS spectra were then acquired on a Bruker Nanostar-U (Bruker, Switzerland) at four selected points distributed around the cross-section. The elastic scattering of x-rays at a nanoscale allows for the evaluation of the relationship between scattered intensity (I), scattering vector (q) and azimuthal angle (ψ). This provides an estimation of thickness, orientation and shape of mineral crystals in bone; the scattering vector q is given by 4sin(θ)/λ (λ is the wavelength of x-rays (0.154 nm) and θ is the angle between the x-ray beam and detector).⁽³³⁾ The thickness of mineral crystals was calculated, following published literature.⁽³³⁾ In brief, the thickness of the mineral particles was based on a calculation of surface to volume ratio with the equation (4 J/πP). Kratky plots (q²I vs. q), were used to evaluate the area under the curve (J), and the Porod constant P was calculated for every 2D spectrum from regions where Iq⁴ was constant, based on Porod’s law (P = Iq⁴). To calculate the orientation of the crystals (ρ), I(ψ) versus ψ (0 < ψ < 2π) was plotted.

Mechanical Characterization

The contributions of HFD and CRD to the biomechanical properties of bone were characterized using strength and fracture mechanics-based tests for toughness. For strength testing, whole tibiae samples were loaded to failure in three-point bending subjecting the anterior surface to tension and the posterior to compression. Samples were tested under wet conditions using a electromechanical testing system (EnduraTEC 3200, TA Instruments, New Castle, DE) at a ramp speed of 0.001mm/s with a span length of 14mm. Load-displacement curves were recorded and converted to stress-strain curves (using the geometry of the mid-diaphysis from the microCT scans) in order to describe whole-bone structural and mechanical properties such as yield strength (MPa), elastic modulus (GPa), ultimate strength (MPa), and post-yield displacement (mm).^(34,35) Yield strength represents the magnitude of loading the bone can resist before permanent deformation begins. The yield point, which marks the point of change from the elastic region to the plastic region, was determined by superimposing a 10% offset line (i.e., 10% loss of stiffness) to the stress-strain curve to find the point where the lines intersected.⁽³⁴⁾ The elastic modulus, or Young’s modulus, was calculated from the slope of the elastic region as a measure of stiffness. Ultimate strength, a measure of the maximum loading the bone can withstand before failure, was directly obtained from the maximum value recorded during the test. The post-yield displacement (PYD) represents the displacement from the yield point to the failure point and was calculated as a measure of ductility. Finally, work-to-fracture (also referred to as energy-to-fracture) is calculated from the area under the load-displacement curve which provides a measure of energy dissipated as bone is resisting fracture.

While strength testing provides load bearing and deformation capacity of bone until fracture, fracture mechanics-based tests provide a measure of fracture toughness or bone’s resistance to fracture due to a pre-existing flaw such as a microcrack or microstructural discontinuities such as lacunae which are inherently present in bone. To obtain resistance to fracture femoral head and condyles were removed from each sample and a flaw, in the form of a controlled notch, was introduced in the mid-diaphyseal region on the anterior surface using a slow-speed diamond saw and sharpened using a razorblade⁽³⁴⁾. Three-dimensional reconstruction of the notch was performed using μCT images to measure the inner radius (Ri), outer radius (Ro), and notch angle with ImageJ. Samples were tested under wet conditions using an electromechanical testing system (EnduraTEC 3200, TA Instruments, New Castle, DE) at a ramp speed of 0.001mm/s with a span length of 4mm. Load and displacement curves were recorded and using published equations^(36,37) we calculated initiation toughness (Kc in; MPa$\surd \mathrm{m}$), maximum toughness (Kc max; MPa$\surd \mathrm{m}$), and the toughening effect (Kc max – Kc in; MPa$\surd \mathrm{m}$). Initiation toughness defines the stress intensity at the initiation of fracture from the notch and the maximum toughness defines the stress intensity at maximum load during crack propagation (commonly referred to as crack growth resistance or propagation toughness). The toughening effect, measured as the difference between maximum and initiation toughness, was calculated to provide a measure of the energy dissipated during fracture.⁽³⁸⁾

Measurement of fAGEs

Total fAGEs in bone samples were measured using protocols previously established in our lab.^(18,19) A section of cortical bone tissue was cut from the mid-diaphysis of each femora sample after mechanical testing. Each sample of approximately 10 mg was washed free of blood with distilled water and subsequently defatted in several cycles of 100% isopropyl ether. All samples were then lyophilized to remove excess liquid from each bone sample. Each bone sample was then placed in a glass vial and submerged with 6N HCl solution at a ratio of 50 μL per 1 mg of bone. These were incubated at 110^o C for direct hydrolysis. Two hydrosylate out of each sample were made by diluting the samples with nanopore water. These were used in two separate assays: a quinine sulphate fluorescence assay, and a collagen content assay.

The first assay used a quinine stock (1 μg Quinine/ mL 0.1 M H₂SO₄) which was diluted with sulfuric acid to create standard curve. Triplicates of each standard and sample hydrolysates were placed into the wells of a 96-microtiter plate. The fluorescence of each sample was measured using a spectrophotometer (Infinite 200, Tecan Trading AG, Switzerland) at 360 nm excitation and 460 nm emission.

The second part of the assay used to a hydroxyproline stock (2000 μg L-hydroxyproline / mL 0.001 M HCl) was diluted with nano-pure water to create a standard curve. Freshly made solutions of chloramine-T, 3.15 M perchloric acid, and p-dimethylaminobenzaldehyde which were added to each standard and sample hydrolysate. After adding the chloramine-T solution, the samples were incubated at room temperature. Next, the 3.15 M perchloric acid solution was added, and the samples were incubated at room temperature. Finally, the p-dimethylaminobenzaldehyde solution was added and the samples were incubated at 60^oC. Triplicates of each standards and sample hydrolysate were placed in a 96-well plate. The absorbance of each sample was measured at 570 nm using the same spectrophotometer. The “in bulk” fAGEs content of each sample was then calculated as a unit of quinine fluorescence normalized by the collagen content (ng quinine/mg collagen).⁽¹⁹⁾

Statistical Analyses

Statistical analyses were performed unblinded and using MATLAB R2020b (The MathWorks, Natick, MA). All data were tested for normality using the Kolmogorov-Smirnov test. In cases where data were non-normal, we performed power transformations, such as the natural log, to normalize the distribution. Bartlett’s test was used to check for homogeneity of variances. Due to sample size being unbalanced (as three animals died prior to culmination of the study and were excluded), we used unbalanced Two-Factor ANOVA with Replication (α < 0.05) to discern the diet, lighting, and interactions effect. Residuals were plotted for normal probability check, and for homogeneity of variances. None of our data violated these assumptions. Post-hocs were conducted using Tukey-Kramer test for multiple comparison. All data is displayed as boxplots (with median and interquartile range) showing all data points. Whole-bone mechanical properties, morphology, and BMD results were adjusted for body mass using a Linear Regression Method as specified for bone biomechanics.⁽³⁹⁾ The blood glucose levels, and bone matrix parameters associated with inorganic (BMD) and organic (fAGEs) matrices, measured in all samples, were subjected to linear regression analyses to determine their contribution to altered bone quality caused by CRD and HFD. Since SAXS parameters were measured in three representative samples, they were not used for correlation. Correlation was determined by Pearson’s R. Power analysis using R software (https://www.rproject.org) demonstrated that the sample size provided sufficient 80% power (α=0.05).

Results

Circadian rhythm disruption in combination with high-fat diet impairs glycemic control

Mice that were fed a HFD displayed a significant increase in body mass when compared to mice fed a LFD (p < 0.0000, Figure 1A). There were no individual light schedule effects present when comparing Dayshift groups to RSS3 groups (p = 0.7005). However, diet and light schedule showed interaction effects (p = 0.0449) to increase body mass as post-hocs revealed RSS3/HFD (diabetic group) was significantly higher compared to Dayshift/LFD (control group) (p-value < 0.0000).

Fig. 1.

(A) Body mass was significantly increased in HFD groups compared to LFD groups. No individual light schedule effects were present, although diet and light schedule displayed interaction effects to increase body mass as RSS3/HFD (diabetic group) was significantly higher compared to Dayshift/LFD (control group) (p-value < 0.0000). (B) Fasting blood glucose levels was significantly increased both by HFD and RSS3 individually; the combination of diet with light schedule also displayed interaction effects as RSS3/HFD was significantly higher compared to Dayshift/LFD (p-value < 0.0000). (C) Area under the curve was increased by both HFD and RSS3 individually. Although ANOVA did not show interaction effects, post-hocs revealed one-way interactions as RSS3/HFD was significantly higher compared to Dayshift/LFD (p-value < 0.0000). LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week; AUC = area under the glucose tolerance test curve. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Fasting blood glucose results showed that only mice fed a HFD while on a shifted light schedule (RSS3/HFD) reached the diabetic range (Figure 1B). Blood glucose levels was significantly increased both by HFD (p < 0.0000) and RSS3 (p = 0.0073) individually; the combination of diet and light schedule also displayed interaction effects (p = 0.0025) as post-hocs identified RSS3/HFD was significantly higher compared to Dayshift/LFD (p-value < 0.0000).

Area under the OGTT curve (AUC) analysis further confirmed the results (Figure 1C). AUC was increased by both HFD (p < 0.0000) and RSS3 (p = 0.0004) individually. Although ANOVA did not show interaction effects (p = 0.2817), post-hocs revealed one-way interactions as RSS3/HFD was significantly higher compared to Dayshift/LFD (p-value < 0.0000) showing that diabetic mice experienced the highest difficulty in restoring the body’s glucose back to their baseline levels, independent of the starting point.

Circadian rhythm disruption in combination with high-fat diet alters cortical bone geometry of femora

Endosteal diameter (Figure 2A) showed both individual diet (p < 0.0000) and light schedule (p = 0.0416) effects, as well as interaction effects between diet and light schedule (p = 0.0003) from ANOVA results. Post-hocs showed RSS3 alone significantly increased (p = 0.0005) endosteal diameter (as compared between RSS3/LFD and Dayshift/LFD). Diet effects were only found within the RSS3 groups, where HFD diminished the increase in endosteal diameter seen with RSS3, as RSS3/LFD and RSS3/HFD showed statistically significant differences (p < 0.0000). Post-hocs showed that interaction effects were between RSS3/LFD and Dayshift/HFD (p = 0.0001).

Fig. 2.

Bone geometry of femora cortical region. (A) Endosteal diameter showed individual diet (p < 0.0000) and light schedule (p = 0.0416) effects, as well as interaction effects between diet and light schedule (p = 0.0003). RSS3 alone significantly increased (p = 0.0005) endosteal diameter (as compared between RSS3/LFD and Dayshift/LFD). Diet effects were only found within the RSS3 groups, between RSS3/LFD and RSS3/HFD (p < 0.0000). Interaction effects were between RSS3/LFD and Dayshift/HFD (p = 0.0001). (B) Periosteal diameter showed both individual diet (p < 0.0000) and light schedule (p = 0.0018) effects, as well as interaction effects between diet and light schedule (p = 0.0014). RSS3 increased (p = 0.0001) periosteal diameter (between RSS3/LFD and Dayshift/LFD). Diet effects were only found within the RSS3 groups, seen with RSS3 as RSS3/LFD and RSS3/HFD showed statistically significant differences (p < 0.0000). Interaction effects were between RSS3/LFD and Dayshift/HFD (p < 0.0000). (C) Cortical thickness (Ct.Th) showed individual diet (p = 0.0005) and light schedule effects (p = 0.0002), but no interaction effects (p = 0.5459). Post-hocs demonstrated that HFD reduced cortical thickness within Dayshift groups (0.00972) and RSS3 groups (p = 0.0214). Light schedule effects were also present with post-hocs showing that RSS3 alone increased cortical thickness (p = 0.007), when comparing RSS3/LFD to Dayshift/LFD. (D) Moment of inertia (Ixx) was reduced with RSS3 groups. Ixx was not affected by diet (p = 0.6187), nor interaction effects (p = 0.4259). RSS3/HFD (diabetic group) reduced Ixx when compared to Dayshift/LFD (control group). RSS3 alone also reduced Ixx when comparing RSS3/LFD to Dayshift/LFD (p = 0.0006). (E) Bone mineral density (BMD) increased between Dayshift/HFD to Dayshift/LFD (p = 0.0001). RSS3 also increased BMD, between RSS3/LFD and Dayshift/LFD (p < 0.0000). Diet and light schedule showed interaction effects (p < 0.0000). Despite individually both HFD and RSS3 increasing BMD, combination of both ultimately reduced mineralization between RSS3/HFD to RSS3/LFD (p < 0.0000) and Dayshift/HFD (p = 0.0003). All normalized by body mass. LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Periosteal diameter (Figure 2B) showed both individual diet (p < 0.0000) and light schedule (p = 0.0018) effects, as well as interaction effects between diet and light schedule (p = 0.0014) from ANOVA results. Post-hocs showed RSS3 alone significantly increased (p = 0.0001) periosteal diameter (as compared between RSS3/LFD and Dayshift/LFD). Diet effects were only found within the RSS3 groups, where HFD diminished the increase in periosteal diameter seen with RSS3 as RSS3/LFD and RSS3/HFD showed statistically significant differences (p < 0.0000). Post-hocs showed that RSS3/LFD had a significant increase in periosteal diameter compared to Dayshift/HFD (p < 0.0000).

Cortical thickness (Ct. Th) (Figure 2C) showed both individual diet (p = 0.0005) and light schedule effects (p = 0.0002), but no interaction effects (p = 0.5459) between factors as revealed by ANOVA. In terms of diet effects, post-hocs demonstrated that HFD reduced cortical thickness within Dayshift groups (0.00972) and RSS3 groups (p = 0.0214). Light schedule effects were also present with post-hocs showing that RSS3 alone increased cortical thickness (p = 0.007), when comparing RSS3/LFD to Dayshift/LFD.

Moment of inertia (I_xx) (Figure 2D) showed reductions with RSS3 groups. I_xx was not affected by diet (p = 0.6187), nor interaction effects between diet and light schedule (p = 0.4259). However, post-hocs tests revealed RSS3/HFD (diabetic group) significantly reduced moment of inertia when compared to Dayshift/LFD (control group). RSS3 alone also reduced I_xx when comparing RSS3/LFD to Dayshift/LFD (p = 0.0006).

Cortical bone mineral density (BMD) (Figure 2E) was significantly increased with HFD within dayshift groups, as compared between Dayshift/HFD to Dayshift/LFD (p = 0.0001). RSS3 alone also increased cortical BMD, as compared between RSS3/LFD and Dayshift/LFD (p < 0.0000). Diet and light schedule showed interaction effects (p < 0.0000). Despite individually both HFD and RSS3 increasing cortical BMD, combination of both ultimately reduced mineralization as comparing RSS3/HFD to RSS3/LFD (p < 0.0000) and Dayshift/HFD (p = 0.0003).

Combination of a high-fat diet with a reversed shift schedule degrades trabecular morphology

Trabecular number (Figure 3A) showed diet effects (p = 0.0009) as HFD reduced trabeculae in both dayshift groups (p = 0.0473) and RSS3 groups (p = 0.0343). There were no light schedule effects (p = 0.2262) nor interaction effects (p = 0.5711) present. Post-hocs revealed RSS3/LFD to have higher trabecular number when compared to Dayshift/HFD (p = 0.0071).

Fig. 3.

Bone geometry of tibiae. (A) Trabecular number (Tb.N) showed diet effects (p = 0.0009) as HFD reduced Tb.N in both dayshift groups (p = 0.0473) and RSS3 groups (p = 0.0343). No light schedule effects (p = 0.2262) nor interaction effects (p = 0.5711) were present. However, post-hocs revealed RSS3/LFD to have higher Tb.N compared to Dayshift/HFD (p = 0.0071). (B) Trabecular separation (Tb.Sp) showed diet effects (p = 0.0015) as HFD increased Tb.Sp in RSS3 groups (p = 0.0079), and dayshift groups (p = 0.0507). No light schedule effects (p = 0.0991) nor interaction effects (p = 0.8057) were present. However, post-hocs revealed RSS3/LFD to have lower Tb.Sp compared to Dayshift/HFD (p = 0.0039). (C) Trabecular thickness (Tb.Th) showed diet effects (p = 0.0198) as RSS3/HFD had increased Tb.Th compared to RSS3/LFD (p = 0.0019). Conversely, RSS3 reduced Tb.Th as compared between RSS3/LFD to Dayshift/LFD (p = 0.0221). Tb.Th showed interaction effects (p = 0.0021) as the RSS3/HFD group (diabetic group) had higher Tb.Th compared to both Dayshift/HFD (p = 0.0424) and Dayshift/LFD (control group) (p = 0.0417). (D) Trabecular bone mineral density (BMD) showed no diet (p = 0.7458), light schedule (p = 0.1178) or interaction effects (p = 0.4806). (E) Bone volume fraction (BV/TV) showed no diet (p = 0.7596), light schedule (p = 0.9584) or interaction effects (p = 0.1545). All normalized by body mass. LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Trabecular separation (Figure 3B) showed diet effects (p = 0.0015) as HFD increased separation between trabeculae in RSS3 groups (p = 0.0079), and dayshift groups (p = 0.0507). No light schedule effects (p = 0.0991) nor interaction effects (p = 0.8057) were present. Post-hocs revealed RSS3/LFD to have lower trabecular separation compared to Dayshift/HFD (p = 0.0039).

Trabecular thickness (Figure 3C) showed diet effects (p = 0.0198) as RSS3/HFD had increased thickness compared to RSS3/LFD (p = 0.0019). Conversely, RSS3/LFD showed reduced thickness compared to Dayshift/LFD (p = 0.0221). Interaction effects (p = 0.0021) were present as the RSS3/HFD group (diabetic group) had higher thickness compared to Dayshift/HFD (p = 0.0424) and Dayshift/LFD (control group) (p = 0.0417).

Trabecular bone mineral density (Figure 3D) showed no diet (p = 0.7458), light schedule (p = 0.1178) or interaction effects (p = 0.4806). Bone volume fraction (Figure 4E) showed no diet (p = 0.7596), light schedule (p = 0.9584) or interaction effects (p = 0.1545).

Fig. 4.

Bone strength measured through three-point bending tests of tibiae. (A) Elastic modulus showed diet (p = 0.0367) and light schedule effects (p = 0.0064), but no interaction effects between them (p = 0.4764). Post-hocs showed HFD increased the elastic modulus within RSS3 groups (p = 0.0499), and the diabetic group (RSS3/HFD) had increased stiffness as compared to control group (Dayshift/LFD) (p = 0.0054). (B) Yield strength showed diet (p = 0.0044) effects, but no light schedule (p = 0.0509) or interaction effects (p = 0.8393). Post-hocs showed HFD increased yield strength within dayshift groups (p = 0.0231) and within RSS3 groups (0.0282). RSS3/HFD group had higher yield strength compared to Dayshift/LFD (p = 0.0001). (C) Ultimate strength showed no diet effects (p = 0.9273). Light schedule (p = 0.0047) and interaction effects (p = 0.0025) were present. Post-hocs revealed HFD increased ultimate strength within dayshift groups (p = 0.0164), while this individual effect was not present in RSS3 groups (p = 0.1563). RSS3 also increased ultimate strength as seen in the increase between RSS3/LFD and Dayshift/LFD (p = 0.0003). When HFD and RSS3 combined, their individual effects mitigate each other and ultimately reduce strength in the diabetic group (RSS3/HFD) as compared to the control group (Dayshift/LFD) (p = 0.0388). (D) Post-yield displacement (PYD) showed diet effects (p = 0.0321) as HFD lowered PYD between Dayshift/HFD and Dayshift/LFD (p = 0.0388). No individual light schedule effects were present (p = 0.7968). Interaction effects between diet and light schedule present (p = 0.0483) as RSS3/HFD (diabetic group) showed reduction in post-yield deformation compared to Dayshift/LFD (control group) (p = 0.0089). (E) Work-to-fracture (Wf) showed diet (p = 0.0008), light schedule (p < 0.0000), and interaction effects (p = 0.0318). RSS3/HFD (diabetic group) displayed lower Wf compared to Dayshift/LFD (control group) (p < 0.0000) and Dayshift/HFD (p < 0.0000). HFD lowered Wf within RSS3 groups, between RSS3/HFD and RSS3/LFD (p = 0.0014). LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Circadian rhythm disruption in combination with HFD increased bone brittleness while leading to loss of fracture resistance

Elastic modulus (Figure 4A) showed diet (p = 0.0367) and light schedule effects (p = 0.0064), but no interaction effects between them (p = 0.4764). Post-hocs showed HFD increased the elastic modulus within RSS3 groups (p = 0.0499), and combination of diet and light schedule (RSS3/HFD) increased modulus compared to control group (Dayshift/LFD) (p = 0.0054).

Yield strength (Figure 4B) showed diet (p = 0.0044) effects, but no light schedule (p = 0.0509) or interaction effects (p = 0.8393). Post-hocs showed HFD increased yield strength within dayshift groups (p = 0.0231) and within RSS3 groups (0.0282). RSS3/HFD group had significantly higher yield strength compared to Dayshift/LFD (p = 0.0001)

Ultimate strength (Figure 4C) showed no diet effects (p = 0.9273). However, light schedule effects (p = 0.0047) and interaction effects between diet and light schedule (p = 0.0025) were present. Post-hocs revealed HFD increased ultimate strength within dayshift groups (p = 0.0164), while this individual effect was not present in RSS3 groups (p = 0.1563). RSS3 also increased ultimate strength as seen in the increase between RSS3/LFD and Dayshift/LFD (p = 0.0003). However, when HFD and RSS3 combined, their individual effects mitigate each other and ultimately reduce strength in the diabetic group (RSS3/HFD) as compared to the control group (Dayshift/LFD) (p = 0.0388).

Post-yield displacement (PYD) (Figure 4D) showed diet effects (p = 0.0321) as HFD lowered PYD between Dayshift/HFD and Dayshift/LFD (p = 0.0388). No individual light schedule effects were present (p = 0.7968). However, interaction effects between diet and light schedule present (p = 0.0483) as RSS3/HFD (diabetic group) showed reduction in post-yield deformation compared to Dayshift/LFD (control group) (p = 0.0089).

The combination of yielding in lower loads with larger deformation is associated with high toughness. In contrast, brittle materials yield at higher forces and display less deformation before failure. Our results showed that HFD had an effect in increasing yield strength (Figure 4B) and lowering PYD (Figure 4D). We also saw an interaction effect (between diet and light schedule) as post-hoc test revealed differences between RSS3/HFD (diabetic group) and Dayshift/LFD (control group) in which the diabetic group displayed brittle behavior.

Furthermore, Work-to-Fracture (W_f) (Figure 4E) showed diet (p = 0.0008), light schedule (p < 0.0000), and interaction effects (p = 0.0318). RSS3/HFD (diabetic group) displayed lower W_f compared to Dayshift/LFD (control group) (p < 0.0000) and Dayshift/HFD (p < 0.0000). HFD lowered W_f within RSS3 groups, between RSS3/HFD and RSS3/LFD (p = 0.0014).

Crystal thickness decreases with HFD, while CRD increases formation of fAGEs demonstrating different outcomes on the mineral versus organic components of the matrix

Mineral crystal thickness (Figure 5A), measured in femora samples, showed diet effects (p = 0.0036), and no light schedule (p = 0.6663) or interaction effects (p = 0.1699). HFD decreased crystal thickness, as compared between Dayshift/HFD to Dayshift/LFD (p = 0.0074). Interestingly, the combination of HFD with CRDs attenuated this difference. Orientation of the crystals showed a decreasing but not significant trend (Figure 5B) in both HFD and CRDs groups. There were no diet (p = 0.3402), light schedule (p = 0.1719) or interaction effects (p = 0.6205) present.

Fig. 5.

Compositional properties of femora. (A) Mineral crystal thickness showed diet effects (p = 0.0036), and no light schedule (p = 0.6663) or interaction effects (p = 0.1699). HFD decreased crystal thickness, as compared between Dayshift/HFD to Dayshift/LFD (p = 0.0074). (B) Orientation of mineral crystals had no diet (p = 0.3402), light schedule (p = 0.1719) or interaction effects (p = 0.6205). (C) Total fAGEs showed light schedule effects (p = 0.0002), and no diet (p = 0.3685) or interaction effects (p = 0.8785). RSS3 increased fAGEs, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0107), and RSS3/HFD to Dayshift/HFD (p = 0.0306). RSS3/HFD (diabetic group) had increased fAGEs compared to Dayshift/LFD (control group) (p = 0.0209). LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week; fAGEs = fluorescent advance glycation end-products. Crystal thickness and orientation was measured from an average of four points, while fAGEs was measured in triplicates. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Total fAGEs measurements (Figure 5C), from femora, revealed an increase in mean glycation for RSS3 mice on HFD by 105.5% and LFD by 96.5% when CRDs were present. Total fAGEs showed light schedule effects (p = 0.0002), and no diet (p = 0.3685) or interaction effects (p = 0.8785). RSS3 increased fAGEs, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0107), and RSS3/HFD to Dayshift/HFD (p = 0.0306). RSS3/HFD (diabetic group) had increased fAGEs compared to Dayshift/LFD (control group) (p = 0.0209).

Circadian rhythm disruption results in loss of femora fracture toughness regardless of diet

Fracture toughness at initiation (Figure 6A) showed diet (p < 0.0000) and light schedule effects (p = 0.0049). There were no interaction effects (p = 0.2807). HFD increased initiation toughness, regardless of light schedule. Dayshift/HFD had higher initiation toughness compared to Dayshift/LFD (p = 0.0012), while RSS3/HFD also had higher initiation toughness, as compared to RSS3/LFD (p < 0.0000). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0289). RSS3/HFD had higher initiation toughness compared to Dayshift/LFD (p = 0.0161).

Fig. 6.

Bone toughness measured through notched tests of femora. (A) Fracture toughness at initiation (Kc, in) showed diet (p < 0.0000) and light schedule effects (p = 0.0049). There were no interaction effects (p = 0.2807). HFD increased initiation toughness, regardless of light schedule. Dayshift/HFD had higher initiation toughness compared to Dayshift/LFD (p = 0.0012), while RSS3/HFD also had higher Kc, as compared to RSS3/LFD (p < 0.0000). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0289). RSS3/HFD had higher initiation toughness compared to Dayshift/LFD (p = 0.0161). (B) Maximum fracture toughness (Kc, max) showed diet (p < 0.0000) and light schedule effects (p = 0.0002). There were no interaction effects (p = 0.601). HFD increased maximum toughness, regardless of light schedule. Dayshift/HFD had higher fracture toughness compared to Dayshift/LFD (p = 0.0475), while RSS3/HFD also had higher Kc, max compared to RSS3/LFD (p < 0.0000). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0005). (C) Toughening effect (ΔKc) showed diet (p = 0.0001) and light schedule effects (p = 0.0003). There were no interaction effects present (p = 0.0534). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0007). Within RSS3 groups, RSS3/HFD displayed higher toughening effect compared to RSS3/LFD (p = 0.0009). LFD = low-fat diet; HFD = high-fat diet; RSS3 = reversed shift schedule 3 times per week. Data is shown as boxplots (with median and interquartile range) showing all data points. Statistics performed by Two-Factor ANOVA with Replication (α=0.05) with Tukey-Cramer post-hoc test for multicomparison. Number next to legend indicates sample size.

Maximum fracture toughness (Figure 6B) showed diet (p < 0.0000) and light schedule effects (p = 0.0002). There were no interaction effects (p = 0.601). HFD increased maximum toughness, regardless of light schedule. Dayshift/HFD had higher fracture toughness as compared to Dayshift/LFD (p = 0.0475), while RSS3/HFD also had higher maximum toughness compared to RSS3/LFD (p < 0.0000). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0005).

Toughening effect (ΔKc), shown in Figure 6C, showed diet (p = 0.0001) and light schedule effects (p = 0.0003). There were no interaction effects present (p = 0.0534). RSS3 groups showed reduced toughness when compared to dayshift groups, as compared between RSS3/LFD to Dayshift/LFD (p = 0.0007). Within RSS3 groups, RSS3/HFD displayed higher toughening effect compared to RSS3/LFD (p = 0.0009).

Pearson’s R was used to determine the coefficient of between fracture toughness results (initiation, maximum, and toughening effect) to total fAGEs. No significant correlation was found between fAGEs and the various measures of toughness within individual groups. When pooling all groups together, correlation analyses revealed that fAGEs have a significant relation to initiation toughness (R = −0.3326, p = 0.0293; Figure 7A), maximum toughness (R = −0.4173, p = 0.0053; Figure 7B) and the toughening effect (R = −0.4197, p = 0.0051, Figure 7C). As noted above SASX based mineral characteristics were measured in only three samples per group, hence non-correlational analysis was done. However, in general, decreased crystal thickness in HFD groups was associated with higher measures of initiation and propagation by groups.

Fig. 7.

Correlation between fAGEs and (A) initiation toughness, (B) maximum toughness, and (C) toughening effect. Analyses determined that fAGEs had a significant relation to initiation toughness (R = −0.3326, p = 0.0293), maximum toughness (R = −0.4173, p = 0.0053) and the toughening effect (R = −0.4197, p = 0.0051). Coefficient of correlation determined by Pearson’s R test (α=0.05). Number on top of each plot indicates sample size.

Discussion

While the effects of hyperglycemia on bone quality have been well reported in literature ^{(15, 16, 40, 41)}, this is the first study to combine circadian rhythm disruption with a high-fat diet in a growing mouse model with a focus on bone quality. Generation of transgenic mice models have long been a classic tool in interrogating T2D pathogenesis because they may provide a direct link between causality of the disease and genetical background.^{(31, 42)} However, recent studies have revealed that environmental factors such as diet and circadian rhythms are also crucial in the development of many diseases such as diabetes, Alzheimer’s disease.^(42–47) Here, we show that circadian rhythm disruption together with a HFD during development can impair glycemic control and have complex effects on bone quality.

While high-fat diets have generally been shown to increase levels of blood glucose in mice⁽⁴⁹⁾, the variety of available nutritional compositions and manufacturing process for such diets can, at times, lead to opposite outcomes.⁽⁵⁰⁾ Recent results highlight the need for additional interventions, such as streptozotocin (STZ) injections with HFD in mice to induce a diabetic skeletal phenotype.⁽⁵¹⁾ Consistent with the above proposition, our study demonstrated that the combination of circadian rhythm disruption and a HFD leads to hyperglycemia. In fact, it was only our combined group (RSS3/HFD) that reached diabetic levels as determined by fasting blood glucose measurements. Moreover, we performed OGTT which allowed us to analyze the area under the curve (AUC) as an index of glucose intolerance.⁽³⁰⁾ We observed significant increase in glucose excursion with HFD and CRD, further confirming that both diet and circadian disruption contribute to increased risk of hyperglycemia. Furthermore, high-fat diets have been shown to increase serum leptin concentrations, which are associated with risk of obesity, and serum Insulin-like Growth Factor 1 (IGF-1) levels that are known play a role in diabetes development.⁽⁴⁹⁾ While we did not measure body composition or serum hormones, the metabolic effects of HFD and CRD have been reported.^{(49, 54, 62)} In agreement, we observed HFD fed mice increased body mass and blood glucose levels. However, circadian disruption had no such effect of body mass as RSS3 groups were stagnantly lean. Nevertheless, circadian disruption impaired glucose tolerance suggesting a markedly different pathway for such occurrence. It has been shown that core clock genes (e.g., CLOCK/BMAL1) found within β cells can influence insulin secretion⁽¹⁰⁾ and disturbances in the sleep/wake cycle cause a pro-inflammatory repose that increases oxidative stress⁽⁵²⁾ presenting a risk for diabetes.⁽¹¹⁾ In concert with these studies, our results show that CRD increased glucose excursion (i.e., elevated AUC) and demonstrated an additive effect with HFD to cause hyperglycemia associated with diabetes.

Our results indicate that an interaction between diet and a shifted light schedule altered bone structure, assessed through microCT, in femora cortical tissue. We found that circadian rhythm disruption increased inner and outer surface expansion of the midshaft. However, this effect was blunted when CRD was combined with a HFD. Moreover, CRD increased femora cortical thickness whereas HFD decreased it. Further, HFD alone led to a significant increase in cortical BMD. It is likely that HFD increased cortical BMD as a result of weight gain, and consequently increased loading. Indeed, our results show that body mass and cortical BMD were correlated in the Dayshift/HFD group. In agreement, Ionova-Martin et al.⁽⁴⁹⁾ demonstrates that mice fed a HFD during growth show reduced endocortical resorption and increased bone formation. Meanwhile, RSS3 group also displayed increased cortical BMD when on a LFD. Although, this increase in BMD was not due to increase body mass as the two variables were poorly correlated in the RSS3 group. While we did not measure bone turnover markers, recent work by Schilperoort et al.⁽⁵⁴⁾ shows that both clock and bone-related genes are rhythmically expressed in bone. Interestingly, they showed that APOE*3-Leiden mice fed a western diet and exposed to an alternating light/dark schedule, such as the one used in our work, reduced bone turnover, bone formation and resorption while increasing mineralization. Thus it is likely that CRD and HFD have two distinct effects on skeletal development.^(54,55)

Furthermore, our results also demonstrate that CRD can disproportionally affect cancellous bone tissue.⁽⁵⁴⁾ Within dayshift groups, HFD lowered tibiae trabecular number affecting trabecular integrity, and this impact of HFD on reducing trabecular number was sustained in RSS3 groups. In particular, circadian rhythm disruption with a HFD caused a significant increase in trabecular thickness, suggesting trabecular architecture is adapting to a stronger more plate-like lattice. Consequently, CRDs effects on trabecular bone integrity vary with the use of LFD and HFD. While, at the outset, this compensatory mechanism might prove advantageous to the structural integrity of bone tissue, these morphological changes have been shown to be ultimately deleterious in diabetic patients.^(56,57)

We also observed different outcomes on the mineral versus organic components of the matrix highlighting the depth of effects of the circadian rhythm disruption and HFD in bone quality. In terms of mineral, SAXS analysis of femora cortical bone demonstrated that mineral crystal thickness was driven down by HFD when compared to LFD group. However, CRD had no effect on mineral quality despite showing effects in mineral quantity, as demonstrated by the increase in cortical BMD.

We observed that CRD exacerbated the production of fAGEs that accumulated in the bone matrix. As discussed above, CRD has been shown to increase oxidative stress and may lead to the formation of reactive oxygen species (ROS).^(52,53) The presence of elevated levels of ROS can, for example, damage structural and enzymatic proteins and facilitate the formation of a variety of glycation products. ⁽⁵³⁾ Type 1 collagen, a structural protein that makes up to 90% of the total proteins present in the organic compartment of bone, is highly susceptible to glycation-based protein cross-linking.⁽¹⁸⁾ Interestingly, in this model, we found that the accumulation of fAGEs was driven by CRD and not HFD as reported in other diet-based diabetic models.⁽⁴⁹⁾ Moving forward, it will be important to quantify levels the levels of glycoxidative and glycation-based changes separately and gain a mechanistic understanding of how these processes drive increase in the amount and diversity of advanced glycation end-products.

Insights into changes of bone morphology and matrix with diet and CRD allow us to provide a framework for the observed differences in whole-bone mechanical properties. Strength testing of the tibiae demonstrated that combination of CRD with HFD increased bone modulus and decreased post-yield deformation (PYD) indicating a reduction in the ability of bone to deform before failure. Next, the combination of yielding at higher forces and displaying less deformation, seen in this group, is associated with brittle materials. Interestingly, HFD fed mice exhibited significantly higher ultimate strength, yet when observed in tandem with a reversed light schedule this increase was mitigated. Finally, consistent with changes in mechanical properties noted above, we observed that combination of CRD with HFD significantly reduced whole-bone toughness as seen by the reduction in the work-to-fracture (W_f). Work-to-fracture takes into account both inorganic and organic components of bone as it considers the complete area under the load-displacement curve thus representing the amount of work needed for bone fracture. Overall, our results from three-point bending of the tibiae demonstrated that our diabetic group displayed lowered post-yield deflection, work-to-fracture, and higher elastic modulus (i.e., stiffness). Furthermore, mechanical results from fracture toughness tests on femora showed that consistent with high fAGEs various measures of toughness were lowered in RSS3 groups. Moreover, consistent with previous results on impact of fAGEs on bone fragility^(18–27), we observed that fAGEs correlated with increased fracture risk when analyzing all experimental groups together. Differences between strength and fracture toughness test may represent different contributions of structural and material properties of bone to energy dissipation and its modification by environmental factors such as CRD and HFD. Collectively, these results demonstrate that combination of CRD with HFD lowers various measures of bone fragility.

Interestingly, we also observed that HFD fed mice during growth maintained higher resistance to fracture when compared to those that were fed a LFD. These results are different from a previous study⁽⁴⁹⁾ showing that fracture toughness, determined for bones of young mice (3-week-old) fed for 16 weeks a HFD (60% kcal fat, 20% kcal carbs, 20% kcal protein), showed no difference when compared to the LFD fed mice while adolescent mice (15-week-old) fed a HFD for the same duration of time displayed lower fracture toughness. These differences could be explained based on age of animals and the diet used. In particular, we used 10-week-old mice, kept the animals on diet for a longer period of 22 weeks, and provided the mice with a HFD that was lower in kcal from fat (46% kcal fat, 36% kcal carbs, 18% kcal protein). Moreover, our analysis of the mineral crystal size in HFD groups revealed that the crystal size was smaller as compared to the femoral cortical bone of LFD groups. Therefore, we propose that the higher fracture toughness can be explained on the basis of altered mineralization, because higher fracture toughness caused by altered mineralization was also reported by others.⁽⁵⁶⁾ It is also likely that the LFD fed mice experienced less remodeling. This conclusion is supported by the fact that the larger mineral crystal size allows easier crack initiation⁽⁵⁷⁾ and lower fracture toughness – and this is what we observed in our study. It is noteworthy that while HFD elevated fracture toughness properties over LFD, it was ultimately CRDs that impaired the aforementioned properties.

The design of our experiments, in particular, the diet and light conditions, should be taken into account when interpreting the results. The amount of fat in our selected HFD extracts less calories from fat than commonly used HFDs. This may have resulted in continued growth and matrix renewal (lack of AGE accumulation) in 10-week-old mice, when the diet was started. The nutritional profile, particularly in terms of vitamin content between diets, was also different for the high-fat diet (4.7 IU/g Vitamin A; 1.2 IU/g Vitamin D-3; 60.6 IU/kg Vitamin E; 0.59 ppm Vitamin K) and low-fat diet (3.8 IU/g Vitamin A; 0.9 IU/g Vitamin D-3; 49.3 IU/kg Vitamin E; 0.48 ppm Vitamin K). Furthermore, the percentages of polyunsaturated fatty acids (PUFAs) needed for the solubilization of fat-soluble vitamins (Vitamins. A, D, E, and K) were reduced with the LFD (1.59% PUFAs) compared to HFD (3.83% PUFAs). These factors could have potentially limited calcium absorption, and consequently reduced bone growth and mineralization.

During the time from adolescence to adulthood, mice experience changes in bone in response to HFD.⁽⁴⁹⁾ Exercise can have an impact in mitigating or confounding these effects. Therefore, we did not include running wheels in mice cages. However, such experimental design did not allow for the measurement of cage activity. Thus, it is unclear whether the altered sleep schedule (RSS3) or regular schedule (dayshift) led to increased exercise. Circadian rhythm disruption in mice has been studied in terms of glucose tolerance⁽²⁹⁾, however it is unclear if the LED green light used in this study design (placed individually in cages) could have resulted in less phase shifting and possibly improved circadian regulation compared to commonly used fluorescent white lighting.

Despite the aforementioned, our study demonstrates that a combination of the altered light/dark schedule with a HFD poses a risk factor for development of hyperglycemia and subsequent poor bone quality. Our results also highlight the complex effects of these environmental factors on bone quality during growth. Of note, it is estimated that 40 to 70% of adults suffer from sleep disturbances and/or disorders. These problems increase with age, for example, among Alzheimer’s disease patients, those suffering from glaucoma, and shift workers.^{(48, 58–61)} Meanwhile, there has also been an increasing number of young adults experiencing sleep disorders.⁽⁴⁸⁾ In addition, childhood and adolescent diabetes is a growing worldwide epidemic. Our study demonstrated that the combination of a disrupted sleep cycle in combination with a diet high in fat ultimately leads to the development of Type 2 Diabetes and deterioration of bone quality.

Nonstandard Abbreviations

CRD

circadian rhythm disruption

NEG

non-enzymatic glycation

AGEs

advanced glycation end-products

fAGEs

fluorescent advanced glycation end-products

RSS3

reversed shift schedule 3 times per week

Kc in

initiation toughness

Kc max

maximum toughness