Deletion of FoxO1, 3, and 4 in Osteoblast Progenitors Attenuates the Loss of Cancellous Bone Mass in a Mouse Model of Type 1 Diabetes

Abstract

Type 1 diabetes is associated with osteopenia and increased fragility fractures, attributed to reduced bone formation. However, the molecular mechanisms mediating these effects remain unknown. Insulin promotes osteoblast formation and inhibits the activity of the FoxO transcription factors. FoxOs, on the other hand, inhibit osteoprogenitor proliferation and bone formation. Here, we investigated whether FoxOs play a role in the low bone mass associated with type 1 diabetes, using mice lacking FoxO1, 3, and 4 in osteoprogenitor cells (FoxO1,3,4^ΔOsx1-Cre). Streptozotocin-induced diabetes caused a reduction in bone mass and strength in FoxO-intact mice. In contrast, cancellous bone was unaffected in diabetic FoxO1,3,4^ΔOsx1-Cre mice. The low bone mass in the FoxO-intact diabetic mice was associated with decreased osteoblast number and bone formation, as well as decreased expression of the anti-osteoclastogenic cytokine osteoprotegerin (OPG) and increased osteoclast number. FoxO deficiency did not alter the effects of diabetes on bone formation; however, it did prevent the decrease in OPG and the increase in osteoclast number. Addition of high glucose to osteoblastic cell cultures decreased OPG mRNA, indicating that hyperglycemia in and of itself contributes to diabetic bone loss. Taken together, these results suggest that FoxOs exacerbate the loss of cancellous bone mass associated with type 1 diabetes and that inactivation of FoxOs might ameliorate the adverse effects of insulin deficiency.

Introduction

Low bone mass and delayed fracture healing are among the multiple complications in patients with type 1 diabetes.^(1–4) Indeed, type 1 diabetes causes a reduction in the accrual of bone mass in children and adolescents,⁽⁵⁾ as well as an 8- to 18-fold increased risk of fragility fractures in both men and women.^(6–8) Rodent models with genetically or pharmacologically induced type 1 diabetes recapitulate the low bone mass and strength observed in humans.⁽⁹⁾ These skeletal defects are associated with reduced serum levels of osteocalcin, a protein secreted by osteoblasts, as well as with a severe reduction in osteoblast number and bone formation rate.^(10–12) In contrast to the well-established deleterious effect on bone formation, the effects of type 1 diabetes on bone resorption are variable, with different studies describing that resorption is increased, decreased, or unaffected depending on the duration of diabetes.^(12–14) Several mechanisms, including impaired osteoblast differentiation, apoptosis, osmotic stress, and inflammation, have been proposed to explain the loss of bone mass in type 1 diabetes.⁽⁴⁾ Nonetheless, the cellular and molecular mechanisms that mediate diabetic osteopenia remain unclear.

The skeletal defects in type 1 diabetes are ameliorated by insulin administration.^(11,15) Insulin promotes proliferation and increases alkaline phosphatase activity and collagen synthesis in osteoblastic cells via direct actions mediated by the insulin receptor.^(16,17) Accordingly, mice lacking the insulin receptor in cells of osteoblast lineage have decreased bone mass and strength as well as low osteoblast numbers, indicating that insulin has osteogenic properties.^(18,19) This evidence strongly suggests that the skeletal defects in type 1 diabetes are attributable to the lack of insulin action in osteoblast lineage cells.

The FoxO family of transcription factors is a critical target of insulin in liver, muscle, and adipose tissue. FoxO1, FoxO3, and FoxO4 recognize the same DNA target sequence⁽²⁰⁾ and are ubiquitously expressed in mammalian tissues, including bone.^(21,22) Although individual FoxOs exert some non-redundant functions, numerous studies in mice with single or combined FoxO deletion in whole body or target tissues have elucidated that FoxO1, 3, and 4 act redundantly to maintain stem cells, suppress tumors, and regulate glucose metabolism.^(23–27) FoxO activity is controlled by post-translational modifications such as phosphorylation and acetylation, which alter FoxO’s subcellular localization, protein stability, and DNA binding. Insulin action via the PI3K/Akt axis phosphorylates FoxO1, 3, and 4 and, thereby, promotes FoxO’s exclusion from the nucleus. In contrast, lack of insulin signaling causes FoxO translocation into the nucleus and stimulates transcription of genes implicated in cell cycle arrest, survival, and stress resistance. Accordingly, FoxOs contribute to the pathophysiology of diabetes by promoting hepatic insulin resistance and beta cell failure.⁽²⁸⁾

We have previously reported that mice with deletion of FoxO1, 3, and 4 in osteoblast progenitors have high bone mass caused by increased osteoblast number and bone formation.⁽²⁹⁾ The anti-osteogenic actions of FoxOs result from binding of FoxOs to β-catenin and prevention of the association between β-catenin and TCF/Lef transcription factors and, thereby, Wnt/β-catenin transcriptional activity. Wnt signaling is stimulated by several extracellular Wnt ligands that bind to Frizzled receptors and recruit the LRP5/6 co-receptors to promote the accumulation and nuclear translocation of β-catenin.⁽³⁰⁾ Wnt/β-catenin signaling in osteoprogenitor cells expressing Osx1 is indispensable for osteoblast differentiation and bone formation.⁽³¹⁾ Furthermore, FoxOs in osteoblasts indirectly inhibit bone resorption by stimulating the expression of osteoprotegerin (OPG), the decoy receptor for receptor activator of NF-κB ligand (RANKL).^(29,32) Here, we sought to determine whether activation of FoxOs in the setting of insulin deficiency mediates the deleterious effects of type 1 diabetes on bone. Our results demonstrate that streptozotocin (STZ)-induced type 1 diabetes in mice suppresses bone formation independently of FoxOs. Nevertheless, deletion of FoxOs prevented the adverse effects of diabetes on bone mass.

Materials and Methods

Animal experimentation

The FoxO1,3,4 ^ΔOsx1-Cre and their littermate FoxO1,3,4^f/f controls were generated by crossing FoxO1,3,4^f/f mice (mixture of FVBn and 129Sv) with hemizygous Osx1-cre transgenic mice using a two-step breeding strategy described previously.⁽²⁹⁾ Genotypes of the offspring were determined by PCR using primers specific for Cre 5′-GCGGTCTGGCAGTAAAAACTATC-3′ and 5′-GTGAAACAGCATTGCTGTCACTT-3′, product size 102 bp, and that detected the wild-type and floxed FoxO1, 3, and 4 alleles.⁽²²⁾ The Osx1-Cre mice were crossed with C57BL/6 wild-type mice to generate Creþ transgenics and their wild-type control littermates. For glucose tolerance test (GTT), glucose (2 g/kg body weight) was injected i.p. to mice fasted overnight. Blood from the clipped mouse tails was collected at indicated times and blood glucose was monitored using blood glucose strips and glucometer (Accu-Check instant, Roche Molecular Biochemicals Corp., Indianapolis, IN, USA). Serum was collected by retro orbital bleeding from mice and insulin was quantified measured using a mouse ultrasensitive ELISA kit obtained by ALPCO (Salem, NH, USA).

To induce type 1 diabetes, 4-week-old control and CKO male mice were injected i.p. with STZ (40 μg/g body weight dissolved in 0.1 M citrate buffer, pH 4.5) for 5 days. STZ in solution is highly labile, therefore STZ solutions were used within 10 minutes of being prepared. The multiple low-dose STZ regime used in this study is adequate to induce hyperglycemia, weight loss, and bone loss without affecting liver and kidney functions.⁽³³⁾ Mice from both genotypes were injected with buffer only as treatment controls. Mice were allocated vehicle or STZ treatment after being stratified by the spine bone mineral density (BMD) at age 4 weeks. Specifically, within each genotype, mice were sorted from low to high BMD values. Mice were then attributed the number 1, 2, 2, 1 successively. All animals of the same number were assigned to the same group. Spine BMD were calculated and compared by t test to assure that means were similar. Spine BMD means and standard deviation for vehicle and STZ-allocated control mice were 0.036±0.008 and 0.035±0.008, respectively, before administering the injections. Spine BMD values of CKO mice assigned vehicle and STZ treatment were 0.034±0.007 days and 0.044±0.01, respectively. Seven after the last injection, non-fasting blood glucose measurements were performed using blood obtained from the mouse tail as described above. Mice with blood glucose levels greater than 250 mg/dL were considered diabetic and included in the study. The cage bedding and water supply of the diabetic mice were frequently replaced and animals were monitored for any signs of distress. All animals were euthanized 40 days after the first injection. To quantify bone formation, mice were injected with tetracycline (15 mg/kg body weight) 8 and 4 days before euthanasia. Protocols involving genetically modified mice and their wild-type littermates were approved by the Institutional Animal Care and Use Committees of the University of Arkansas for Medical Sciences and the Central Arkansas Veterans Healthcare System.

Micro-computed tomography (μCT), biomechanical testing, and histomorphometry

Femur and vertebra (L₅) microarchitecture was analyzed with μCT (model μCT40, Scanco Medical, Bruttisellen, Switzerland). Scans were performed at medium resolution (nominal isotropic voxel size=12 μm) and integrated into 3D voxel images (1024×1024 pixel matrices for each individual planar stack). A Gaussian filter (sigma=0.8, support=1) was applied to all analyzed scans. Key parameters were X-ray tube potential = 55 kVp, X-ray intensity=145 μA, integration time = 200 ms, and threshold=200 mg/cm³. The entire vertebral body was scanned with a transverse orientation excluding any bone outside the vertebral body. In the distal femur, 151 transverse slices were taken from the epicondyles and extending toward the proximal end of the femur. All cancellous measurements were made by manually drawing contours every 10 to 20 slices and using voxel counting for bone volume per tissue volume and sphere filling distance transformation indices without assumptions about the bone shape as a rod or plate for trabecular microarchitecture. This excluded the cortical bone and the primary spongiosa from the cancellous bone analysis. Cortical thickness was determined using 18 transverse slices at the femoral mid-diaphysis. Total and medullary area and circumference measurements were calculated from these slices. Image processing language scripts including the “cl image” command were used to obtain the femoral endocortical and periosteal circumference. Micro-CT measurements were expressed in 3D nomenclature as recommended.^(34,35) Compression testing of L₅ was performed to determine bone strength using a single column material testing machine and a calibrated tension/compression load cell (Model 5542; Instron Corp., Canton, MA, USA), as previously described.⁽³⁶⁾ The lumbar vertebrae (L₁ to L₃) were fixed in 10% Millonig’s formalin, transferred to 100% ethanol, and embedded undecalcified in methyl methacrylate. Histomorphometric examination was performed in longitudinal sections using the OsteoMeasure Analysis System (OsteoMetrics, Inc., Decatur, GA, USA) as previously described.^(37,38) Static and dynamic histomorphometry measurements of the cancellous bone were restricted to the secondary spongiosa. The terminology used in this study has been recommended by the Histomorphometry Nomenclature Committee of the American Society for Bone and Mineral Research.^(34,35)

TaqMan assay

Total RNA was extracted from frozen osteocyte-enriched tibia shafts or cultured cells after homogenizing the samples in Trizol Reagent (Life Technologies, Grand Island, NY, USA) according to the manufacturer’s instructions. The mRNA was reverse-transcribed to cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA). The cDNA was amplified by quantitative RT-PCR using TaqMan Universal PCRMaster Mix (Life Technologies) with primers and probes for different genes manufactured by Taqman Gene Expression Assays service (Applied Biosystems) according to the manufacturer’s directions. The following TaqMan assays from Life Technologies were used: osteocalcin (forward 5′GCTGCGCTCTGTCTCTCTGA3′, reverse 5′TGCTTGGACATGAA GGCTTTG3′, probe FAM5′AAGCCCAGCGGCC3′NFQ); osteoprotegerin (OPG) Mm00435452_m1; RANKL Mm00441908_m1; SOST Mm00470479_m1; DKK1 (forward 5′GGGCTGTGTTGTGCAAGACA3′, reverse 5′GGTGCACACCTGACCTTCTTTAA3′, probe FAM 5′TTCTGGTCCAAGATCT3′NFQ); Wnt10b Mm00442104_m1, and the house-keeping gene ribosomal protein S2, Mm00475528_m1. mRNA levels were calculated by normalizing to ribosomal protein S2 using the delta Ct method.

Cell culture

Calvaria cells were isolated from neonatal pups as described before.⁽²⁹⁾ Bone marrow cells were obtained by flushing the femoral diaphysis from 3- to 4-month-old mice with α-MEM medium (Invitrogen, Carlsbad, CA, USA). Cultures of OB-6, an established osteoblastic cell line of bone marrow origin,⁽³⁹⁾ bone marrow stromal, and calvarial cells were maintained in α-MEM containing 10% fetal bovine serum, 1% penicillin/streptomycin/glutamine, and ascorbic acid (50 μg/mL) up to 70% confluence. Calvaria cells were treated with either vehicle, 10 nM insulin (Sigma, St. Louis, MO, USA), 60 mM D-glucose, or 60 mM mannitol for 8 and 24 hours. OB-6 cells were treated with either vehicle or Wnt conditioned medium (1:2 dilution) or elevated (30 mM and 60 mM) D-glucose for 48 hours. Treatment with 30 mM and 60 mM D-mannitol was used as osmolarity controls. OPG expression was measured from RNA isolated from the cell lysates after extraction with Trizol.

ELISA

OPG concentration in bone marrow plasma harvested from long bones was determined using the ELISA assay (MOP00, R&D Systems, Minneapolis, MN, USA) per the manufacturer’s protocol. Bone marrow plasma were collected by removing both ends of femur and tibia with a scalpel and then centrifuging the diaphyseal bone in a microcentrifuge tube at 850g for 30 seconds. The bone marrow cell pellet and bone marrow plasma were then resuspended in 80 μL of phosphate-buffered saline and then centrifuged at 2300g for 1 minute. The supernatant was transferred to a fresh tube and stored at −80°C until analyzed.

Statistics

Data were analyzed using a two-way ANOVA to detect statistically significant treatment effects, after determining that the data were normally distributed and exhibited equivalent variances. In some cases, transformations were used to obtain normally distributed data and equal variance. This was followed by all pairwise comparisons using Tukey’s procedure. For some variables, the p values were adjusted using Holm’s multiple comparison procedure. For experiments involving comparison of only two groups, Student’s t test was used. Any p values less than 0.05 were considered significant. Most data are presented as box plots with the central box spanning 25^th to 75^th percentiles and the central line representing the mean. The whiskers represent 10^th and 90^th percentiles and values outside this range are shown as dots. Data from in vitro studies are plotted as bar graphs that represent the mean and SD.

Results

Deletion of FoxO1, 3, and 4 in osteoblast progenitors does not alter glucose metabolism

To determine whether FoxOs mediate the negative effects of type 1 diabetes on bone in vivo, we used mice with and without FoxO1, 3, and 4 in cells of the osteoblast lineage expressing Osx1-Cre. The Osx1-Cre transgene directs recombination in committed osteoblast progenitors, osteoblasts, osteocytes, hypertrophic chondrocytes, and adipocytes.^(31,40) Four-week-old male FoxO1,3,4^ΔOsx1-Cre mice, hereafter referred to as conditional knockouts (CKO), and their FoxO1,3,4^f/f littermates (controls) were treated with either vehicle or STZ to induce diabetes. About 85% to 90% of the mice injected with STZ had >250 mg/dL blood glucose and were considered diabetic (Fig. 1A). Forty days after the first injection, diabetic and healthy mice of both genotypes were euthanized for further analysis. The diabetic control mice had lower body weight compared with the healthy littermates (Fig. 1B). The CKO mice had lower body weight, caused by the expression of the Osx1-Cre transgene,^(41,42) which was not altered by diabetes. However, the unkempt appearance and the damp cage bedding of diabetic CKO mice was comparable to that of the diabetic control mice.

Fig. 1.

Deletion of FoxOs in Osx1-Cre–expressing cells does not alter glucose metabolism. (A) Blood glucose of the 4-week-old male mice 7 days after being injected with vehicle or STZ for 5 consecutive days (n=8–15/group). (B) Body weight of the mice described in A, 40 days after the first injection. (C) Serum insulin levels in 3-month-old male mice at random feeding (n=8–10/group). Data presented as box plots with central box spanning 25^th to 75^th percentiles and the central line is mean. The whiskers represent the 10^th and 90^th percentiles, and values outside of this range are shown as dots.*p < 0.05 versus vehicle-treated mice of the same genotype; #p < 0.05 versus vehicle-treated control mice by two-way ANOVA. (D) Glucose tolerance test in 5 month old male mice (n=5–6/group). Data represent mean±SEM.

Others have suggested that deletion of FoxO1 in osteoblasts expressing CollAgEn1A1 (Col1A1)-CrE increases serum insulin and improves glucose tolerance.⁽⁴³⁾ In contrast to these earlier findings, serum insulin and glucose levels were not affected by FoxO1,3,4 deletion using Osx1-Cre (Fig. 1A, C). To further examine whether deletion of FoxOs per se altered glucose metabolism, we performed a glucose tolerance test (GTT). A similar response to glucose administration was observed in control and CKO mice, indicating that insulin secretion was unaffected by FoxO deletion (Fig. 1D). We also performed GTT on an independently generated cohort of Osx1-Cre mice and respective wild-type littermates. Mice of both genotypes had similar GTT responses, indicating that the Osx1-Cre transgene exerts no effect on glucose absorption (Supplemental Fig. S1). In view of the evidence that Osx1-Cre targets all osteoblasts in bone,⁽⁴⁰⁾ the reasons for the different results obtained between ours and the published studies are unclear. One plausible explanation for the discrepant results is the different genetic background of the mice used in the experiments.

Deletion of FoxOs in osteoblast progenitors abrogates the loss of cancellous bone mass with diabetes

Diabetic control mice had lower cancellous bone volume, as determined by micro-CT in the distal femurs (Fig. 2A). This change was associated with a reduction in trabecular number and an increase in trabecular separation, whereas trabecular thickness was not affected. In contrast, the adverse effects of diabetes on cancellous bone were abrogated in the CKO mice (Fig. 2A). As we have previously shown,⁽²⁹⁾ CKO mice exhibited increased cortical bone mass (Fig. 2B). Mice with diabetes had decreased cortical thickness and cortical area relative to their non-diabetic controls in both genotypes. Total cortical bone area decreases modestly with diabetes in both genotypes; however, these changes were not statistically significant (Fig. 2B). Marrow area was not different between the diabetic and healthy mice in both genotypes.

Fig. 2.

FoxO deletion in osteoblast progenitors prevents the loss of cancellous but not cortical bone in femora of diabetic mice. Micro-CT of femurs from mice injected with vehicle or STZ (n=8–15/group). (A) Cancellous bone volume and microarchitecture at the metaphysis. BV/TV=bone volume per tissue volume; Tb=trabecular. (B) Cortical thickness and bone area, total and medullary areas at the diaphysis.*p < 0.05 versus vehicle-treated mice of the same genotype; #p < 0.05 versus vehicle-treated control mice by two-way ANOVA.

Diabetes also lowered cancellous bone volume and trabecular thickness in the vertebrae of control but not the CKO mice (Fig. 3A). Accordingly, diabetes lowered vertebral strength of the control but not the CKO mice, as determined by compression studies (Fig. 3B). Despite increased cortical thickness under basal conditions in the CKO mice, both control and CKO mice lost cortical bone with diabetes (Fig. 3C).

Fig. 3.

The adverse effects of diabetes on cancellous bone are abrogated by FoxO deletion in osteoprogenitors. (A) Cancellous bone volume and microarchitecture of the lumbar vertebra (L₅) from mice injected with vehicle or STZ (n=8–15/group). BV/TV=bone volume per tissue volume; Tb=trabecular. (B) Strength determined by compression (n=8–11/group). (C) Cortical thickness of the vertebral ventral wall.*p < 0.05 versus vehicle-treated mice of the same genotype; #p < 0.05 versus vehicle-treated control mice by two-way ANOVA.

Type 1 diabetes decreases bone formation independently of FoxOs

We next performed histological analysis of undecalcified vertebral sections to examine the cellular mechanisms responsible for diabetes-induced changes in bone. Type 1 diabetes decreased osteoblast number and surface in cancellous bone of control mice (Fig. 4A). Accordingly, mineralizing (tetracycline labeled) perimeter (M.Pm/B. Pm) as well as mineral apposition rate (MAR), the distance between the tetracycline labels, were lower in diabetic control mice (Fig. 4B, C). These changes led to a 40% reduction in bone formation rate (BFR=MAR×Mn.Pm/P. Pm). The non-diabetic CKO mice had higher osteoblast surface and number, mineralizing surface, and BFR when compared with the non-diabetic controls. Diabetes in CKO mice also caused a reduction in all osteoblast parameters and lowered bone formation rate by 37%. In line with the changes found in osteoblast number, the expression of the osteoblast-specific gene osteocalcin in bone was decreased with diabetes in both control and CKO mice when compared with their respective healthy littermates (Fig. 4D). Notably, although diabetes decreased osteoblast numbers and bone formation in the CKO mice, these parameters remained significantly higher than the ones in the diabetic controls and similar to the healthy control mice.

Fig. 4.

Type 1 diabetes decreases bone formation independent of FoxOs in osteoblast lineage. (A) Osteoblast (Ob) number and surface per mm cancellous bone in vertebral bone sections from mice injected with vehicle or STZ (n=6/group). (B) Representative photomicrographs (scale bar=20 μm) and (C) mineralizing perimeter (M.Pm/B.Pm), mineral apposition (MAR), and bone formation rate (BFR) determined by tetracycline labels in sections described in A. (D) Gene expression determined by qRT-PCR in femoral and tibia shafts (n=8–15/group).*p < 0.05 versus vehicle-treated mice of same genotype and #p < 0.05 versus control mice with same treatment by two-way ANOVA.

Attenuation of Wnt signaling due to upregulation of Wnt signaling inhibitors Sost and DKK1 and decreased expression of Wnt10b have been suggested as potential mechanisms for the decrease in osteoblast number with diabetes.^(44,45) To determine whether changes in these modulators of Wnt signaling could be responsible for the decrease in bone formation with diabetes, independent of FoxOs, we measured Sost, DKK1, and Wnt10b levels in femoral bone. The mRNA expression of Sost and DKK1 were not altered by diabetes in control mice (Fig. 4D). In contrast, Wnt10b was decreased in diabetic control mice. Deletion of FoxOs had no effect on the levels of any of the Wnt modulators in healthy mice but blunted the inhibitory effects of diabetes on Wnt10b.

Type 1 diabetes stimulates resorption and this effect is abrogated by deletion of FoxOs

In addition to low bone formation, the diabetic control mice had higher number of osteoclasts compared with the healthy controls (Fig. 5A). This was associated with a decrease in the expression of OPG mRNA in bone (Fig. 5B) and protein levels in the bone marrow plasma (Fig. 5C). We had previously reported that the healthy CKO mice exhibit increased number of osteoclasts with a concomitant decrease in OPG,⁽²⁹⁾ similar to the findings in this study. The effects of diabetes on OPG and bone resorption were prevented in the CKO mice. Unlike OPG, the mRNA levels of RANKL were similar in mice from both genotypes and unaffected by diabetes (Fig. 5B). Previous studies have suggested that insulin attenuates OPG mRNA in osteoblasts.⁽³²⁾ This evidence, together with our present findings, suggests that factors other than insulin contribute to the decrease in OPG expression in diabetes. Thus, we examined whether hyperglycemia could be responsible for the decreased OPG expression in diabetes. To this end, we cultured OB-6 osteoblastic cells with normal or elevated glucose concentrations for 48 hours. D-mannitol, a sugar with structure similar to glucose, was used at the same concentration as glucose to determine whether changes in osmolarity could alter OPG. Wnt3a, a potent stimulator of OPG, was used as a positive control. As expected, treatment with Wnt3a conditional medium increased OPG mRNA in Ob-6 cells (Fig. 5D). In contrast, elevated glucose, but not mannitol, suppressed OPG expression by 50% in these cells. Likewise, high glucose decreased OPG levels in primary calvaria osteoblast cells after 8 and 24 hours of culture. Insulin suppressed OPG mRNA transiently at 8 hours but not 24 hours (Fig. 5E). To determine whether the effect of glucose on OPG was dependent on FoxOs, we used bone marrow–derived osteoblasts from CKO mice (Supplemental Fig. S2). The inhibitory actions of glucose on OPG mRNA expression observed in cells from control mice were prevented in cells from CKO mice (Fig. 5F). These findings suggest that hyperglycemia could be responsible for the decline in OPG expression with type 1 diabetes in a FoxO-dependent manner.

Fig. 5.

Type 1 diabetes decreases OPG and promotes bone resorption. (A) Osteoclast (Oc) number and perimeter quantified in vertebral sections (L₁ to L₃) from mice injected with vehicle or STZ (n=6/group). (B) mRNA by qRT-PCR of femur and tibia shafts (n=8–15/group).*p < 0.05 versus vehicle-treated mice of the same genotype; #p < 0.05 versus vehicle-treated control mice by two-way ANOVA. (C) Protein levels as determined by ELISA from bone marrow plasma of mice described in B. (D) OB-6 cells (triplicates) cultured with the indicated compounds for 48 hours; (E) calvaria cell cultures (triplicates). (F) Bone marrow stromal cells (triplicates) cultured in medium containing 1% ascorbate were treated with vehicle or 60 mM glucose.*p < 0.05 versus vehicle-treated mice of the same genotype; #p < 0.05 versus vehicle-treated control mice by two-way ANOVA. ≠p < 0.05 versus vehicle by one-way ANOVA.

Discussion

The present study provides evidence that inhibition of FoxOs in osteoblast lineage cells attenuates the cancellous loss of bone mass in mice with STZ-induced insulin deficiency. We found that STZ-induced diabetic mice exhibited low osteoblast number and bone formation rate, in line with previous reports by others.^(11,12) Because of the inhibitory actions of insulin on FoxO activity and the suppressive effects of FoxOs on bone formation, we had hypothesized that FoxO activation in the absence of insulin could mediate the deleterious effects of diabetes on bone. Contrary to our hypothesis, we found that insulin deficiency lowered BFR and osteoblast number both in FoxO-intact and FoxO-deficient mice. These findings indicate that mechanisms unrelated to FoxOs must be responsible for the adverse effect of insulin deficiency on bone formation. Albeit, the FoxO-deficient mice were protected from the loss of bone mass most likely because the diabetes-induced decrease in OPG and increase in osteoclast number was prevented.

Others have suggested that the decreased bone formation in type 1 diabetes is the result of increased oxidative stress.⁽¹⁴⁾ Transgenic mice with global overexpression of thioredoxin-1 (an antioxidant enzyme that facilitates the reduction of substrates through cysteine thiol-disulfide exchange) were resistant to STZ-induced diabetic osteopenia. However, thioredoxin-1 overexpression ameliorated STZ-induced oxidative stress in various organs and attenuated diabetic nephropathy.⁽⁴⁶⁾ Because renal failure can disrupt bone metabolism,⁽⁴⁷⁾ it remains unclear whether the positive effects of thioredoxin-1 on the osteopenia caused by diabetes were secondary to the protective effects in the kidney. A decrease in Wnt signaling may be an alternative culprit of low bone mass in type 1 diabetes. However, in difference to previous results from others,⁽⁴⁴⁾ we found in the present work no changes in the mRNA expression of Sost and DKK1 in bone of diabetic mice; albeit, the mRNA levels of Wnt10b were reduced, in agreement with Zhang and colleagues.⁽⁴⁵⁾ Endogenous Wnt10b stimulates bone formation and increases cancellous bone mass.⁽⁴⁸⁾ We found that the diabetes-induced decrease in Wnt10b was prevented in the FoxO-deficient mice. However, the decline in osteoblasts was not. We have previously shown that FoxOs do not alter Wnt10b mRNA levels in cultured calvaria cells.⁽²⁹⁾ It is, therefore, possible that the changes observed in Wnt10b in our model merely reflect osteoblast numbers or activity. In any case, our findings argue that mechanisms other than Wnt10b suppression cause the decrease in osteoblast number. Alternatively, bone marrow cells other than those of the osteoblast lineage cells may be contributors of Wnt10b in diabetes. Future studies with mice lacking Wnt10b in cells of the osteoblast lineage are needed to elucidate the contribution of Wnt10b to the skeletal effects of diabetes.

STZ administration to FoxO-intact mice caused an increase in osteoclast number and a decrease in OPG. The decrease in OPG was also found in non-diabetic FoxO-deficient mice, in agreement with previous findings, by us and others, that FoxOs stimulate OPG.^(29,32) Insulin decreases OPG and increases osteoclast and bone resorption by inhibiting FoxO-mediated transcription.⁽³²⁾ Based on this evidence, we had expected that insulin deficiency will increase OPG and decrease bone resorption, but this was not the case. Thus, alternative mechanisms must be responsible for the effects of type 1 diabetes on bone resorption. We found that exposure of osteoblastic cells to high glucose in vitro decreases OPG mRNA. In agreement with these findings, methylglyoxal, an intermediate metabolite of glucose that is increased in diabetics,^(49,50) decreases OPG mRNA in an osteoblast progenitor cell line.⁽⁵¹⁾ Moreover, type 1 diabetes exacerbates the severity of periodontal bone loss by lowering OPG and promoting osteoclastogenesis.⁽⁵²⁾ Along with our findings, these observations support the contention that chronic hyperglycemia may promote bone resorption by attenuating OPG.

In the present work, the STZ-induced increase in osteoclast number was abrogated in the FoxO-deficient mice. This and the observation that high glucose decreases OPG expression in a FoxO-dependent manner, suggests that hyperglycemia by itself may suppress FoxOs. Consistent with this, in vitro exposure of macrophages to high glucose reduces FoxO1 expression. Moreover, bone marrow–derived macrophages from obese db/db mice or STZ-treated mice have reduced levels of FoxO1.⁽⁵³⁾ Albeit, others have reported that osteoclast number is unaltered or even decreased with type 1 diabetes.^(9,14) One explanation for these incongruent findings is the genetic background of the mice used, as mouse strain strongly influences the severity of diabetes and insulin resistance.^(54–56) For example, DBA/2J mice are more sensitive to STZ-induced diabetes than C57BL/6, 129/SvEv, or BALB/c.⁽⁵⁴⁾ In addition, DBA/2J mice have higher rates of mortality and are more prone to diabetic nephropathy, whereas C57BL/6J mice survive longer after onset of diabetes and are relatively resistant to the development of diabetic nephropathy. In another study comparing129×1/Sv, C57BL/6, DBA/2, and FVB/N mouse strains, the authors demonstrated that insulin secretion is highest in 129×1/Sv mice and that FVB/N mice have hepatic insulin resistance.⁽⁵⁶⁾ The metabolic changes notwithstanding, Balb/c, C57BL/6, and 129/Sv consistently exhibit STZ-induced loss of bone mass, albeit the magnitude of this effect varies perhaps because of differences in bone turnover rates.⁽⁵⁷⁾

Type 1 diabetes causes loss of bone in both the cancellous and cortical compartments. These changes are prevented by administration of insulin.^(11,15) Nonetheless, these findings have not been reproduced in all the studies.^(58,59) We found here that FoxO deficiency prevented the loss of cancellous, but not cortical, bone in the STZ-treated mice, suggesting that the mechanisms mediating the loss of bone in the two compartments are different. In support for this notion, systemic hormones like sex steroids and Wnt ligands like Wnt5a, Wnt10b, and Wnt16 have different effects on cancellous and cortical bone.^(48,60–62) Specifically, the anti-resorptive effect of estrogens on cancellous bone are mediated via direct actions on osteoclasts, whereas in cortical bone the effects of estrogens are exerted indirectly via cells of the osteoblast lineage. Further, Wnt16 attenuates bone resorption at the endocortical surface but not cancellous bone, whereas Wnt5a and Wnt10b promote bone formation and increase cancellous bone mass but have no effect on cortical bone.

The STZ-induced diabetes model has some limitations. Indeed, STZ suppresses liver NAD content.⁽⁶³⁾ However, this effect is transient and reversed 48 hours after the drug injection. STZ also causes progressive renal dysfunction, but this effect is restored with administration of insulin, indicating that the toxicity is due to diabetic nephropathy.⁽⁶⁴⁾ These effects are found in studies in which a single high dose of STZ was used.⁽⁶⁵⁾ Currently, multiple low doses of STZ are used to induce diabetes in mice. This mode of drug administration elicits minimal hepatic and renal dysfunction.^(66,67) More important, the effects of STZ on cancellous and cortical bone, under this protocol, must be attributable to insulin deficiency because the loss of bone mass is greatly attenuated by insulin replacement. Nevertheless, Motyl and McCabe have reported that even though the bone loss with STZ is reproducible, the magnitude of the effect may vary depending on the activity/strength of the drug.⁽³³⁾

In conclusion, we have found that FoxO deficiency prevents the loss of cancellous bone and vertebral strength in diabetic mice. The effect of FoxO deficiency could not be accounted by a reversal of the suppressive effect of STZ on bone formation. Instead, our results suggest that FoxO deficiency abrogated the STZ-induced loss of cancellous bone by preventing the decrease in OPG. The findings of the present report along with earlier evidence that deletion of FoxO1,3,4 in osteoprogenitors of adult mice increases bone mass suggest that inactivation of FoxOs might represent a therapeutic approach to combat the loss of bone mass in diabetes. Our present studies were conducted in young mice. Type 1 diabetes is typically diagnosed in childhood and has an impact on skeletal integrity and possibly peak bone mass. Thus, this work provides pertinent insights into the mechanisms by which insulin deficiency and downstream mediators, like FoxOs, can impact the growing skeleton. Several natural and synthetic small molecule inhibitors of FoxO1, including AS1842856, improve insulin resistance in mice and human tissues.⁽⁶⁸⁾ Additionally, AS1842856 decreases adipogenesis.⁽⁶⁹⁾ Taken together with the present findings, these pharmacological results raise the possibility that FoxO suppression could ameliorate several complications of diabetes, including osteopenia.