FOXO1 Deletion Reverses the Effect of Diabetic-Induced Impaired Fracture Healing

Abstract

Type 1 diabetes impairs fracture healing. We tested the hypothesis that diabetes affects chondrocytes to impair fracture healing through a mechanism that involves the transcription factor FOXO1. Type 1 diabetes was induced by streptozotocin in mice with FOXO1 deletion in chondrocytes (Col2α1Cre⁺FOXO1^L/L) or littermate controls (Col2α1Cre^-FOXO1^L/L) and closed femoral fractures induced. Diabetic mice had 77% less cartilage and 30% less bone than normoglycemics evaluated histologically and by micro-computed tomography. Both were reversed with lineage-specific FOXO1 ablation. Diabetic mice had a threefold increase in osteoclasts and a two- to threefold increase in RANKL mRNA or RANKL-expressing chondrocytes compared with normoglycemics. Both parameters were rescued by FOXO1 ablation in chondrocytes. Conditions present in diabetes, high glucose (HG), and increased advanced glycation end products (AGEs) stimulated FOXO1 association with the RANKL promoter in vitro, and overexpression of FOXO1 increased RANKL promoter activity in luciferase reporter assays. HG and AGE stimulated FOXO1 nuclear localization, which was reversed by insulin and inhibitors of TLR4, histone deacetylase, nitric oxide, and reactive oxygen species. The results indicate that chondrocytes play a prominent role in diabetes-impaired fracture healing and that high levels of glucose, AGEs, and tumor necrosis factor-α, which are elevated by diabetes, alter RANKL expression in chondrocytes via FOXO1.

Figure 1

Cre recombinase deletes FOXO1 in chondrocytes of Col2α1Cre⁺.FOXO^L/L but not control Col2α1Cre⁻.FOXO^L/L mice. A: Genotyping of experimental and littermate control mice. FOXO1 with flanking loxP sites (149 bp) is detected in all groups, whereas cleaved FOXO1 (198 bp) is found only in experimental mice with Cre recombinase (352 bp). B and C: Photomicrographs of histologic sections from the fracture site were examined by immunofluorescence with antibody specific for FOXO1, and nuclei were identified by DAPI counterstain. No signal was detected in matched control antibody. The white dotted line indicates the interface between cartilage and bone. Original magnification ×100. D: Photomicrographs of primary chondrocytes isolated from experimental Col2α1Cre⁺.FOXO1^L/L and littermate control Col2α1Cre⁻.FOXO1^L/L mice. Immunofluorescence (IF) was carried out with antibody (AB) specific to FOXO1 or control IgG. Magnification ×400. E: Mean fluorescence intensity (MFI) was measured from specimens described in D. F: Primary chondrocytes were isolated from experimental Col2α1Cre⁺.FOXO1^L/L (knockout) and littermate control Col2α1Cre⁻.FOXO1^L/L mice. Immunofluorescence was carried out with antibody specific to collagen-2α1, a chondrocyte marker, or control IgG, and MFI was measured. G: Primary chondrocytes were isolated from the control Col2α1Cre⁻.FOXO1^L/L and experimental Col2α1Cre⁺.FOXO1^L/L mice and grown to confluence. They were then cultured in chondrogenic media for 4 days. RNA was obtained, and FOXO1 mRNA levels were measured by real-time PCR to quantify the knockdown efficacy. Quantitative data are expressed as the mean ± SEM. *Significant difference between specimens from Col2α1Cre⁺.FOXO1^L/L and littermate control Col2α1Cre⁻.FOXO1^L/L mice determined by Student t test (P < 0.05).

Figure 2

Diabetes significantly reduces the amount of cartilage and bone, which is rescued by FOXO1 deletion in chondrocytes. Histologic sections at the fracture site were obtained 16 days after fracture from normoglycemic (NG) Col2α1Cre⁻.FOXO1^L/L mice (A), diabetic (Diab) Col2α1Cre⁻.FOXO1^L/L mice (B), and diabetic (Diab) Col2α1Cre⁺.FOXO1^L/L (C) mice with FOXO1 deletion and stained with Safranin-O/fast green. The callus was defined as the area within the peripheral fibrous capsule, excluding the original cortical bone. Cartilage is stained red. Scale bar = 1 mm. D: Safranin-O/fast green–stained sections from the fracture site were examined for cartilage area 10 and 16 days after fracture, bone area 16 and 22 days after fracture, and total callus size 16 and 22 days after fracture. Data are expressed as mean ± SD; n = 5–7/group. Significance was determined by ANOVA followed by Tukey post hoc test (P < 0.05). *Significant difference between specimens from diabetic and matched NG animals; +significant difference between specimens from diabetic Col2α1Cre⁺.FOXO1^L/L and Col2α1Cre⁻.FOXO1^L/L mice.

Figure 3

Diabetes impairs bone formation during fracture healing, which is rescued by FOXO1 deletion in chondrocytes. Micro-CT was performed on fracture calluses during healing 22 days after fracture in animals from normoglycemic (NG) Col2α1Cre⁻.FOXO1^L/L mice (A), diabetic Col2α1Cre⁻.FOXO1^L/L (B) mice, and diabetic Col2α1Cre⁺.FOXO1^L/L mice (C) with FOXO1 deletion. D: The callus volume and amount of bone volume were quantified in micro-CT images from mice 16 and 22 days after fracture. Data are expressed as mean ± SD; n = 6–8/group. Significance was determined by ANOVA followed by Tukey post hoc test (P < 0.05). *Significant difference between specimens from diabetic and matched NG animals; +significant difference between specimens from diabetic Col2α1Cre⁺.FOXO1^L/L and Col2α1Cre⁻.FOXO1^L/L mice.

Figure 4

FOXO1 deletion in chondrocytes rescues the increase in osteoclasts caused by diabetes. Histologic sections were obtained 10 days after fracture from normoglycemic (NG) Col2α1Cre⁻.FOXO1^L/L mice, diabetic Col2α1Cre⁻.FOXO1^L/L mice, or diabetic Col2α1Cre⁺.FOXO1^L/L mice with FOXO1 deletion. Sections were stained for TRAP. Original magnification ×100 (A) and ×200 (B–D). E: Osteoclasts adjacent to cartilage were counted as multinucleated TRAP-positive cells or cathepsin K–positive cells determined by immunohistochemistry with antibody specific to cathepsin K (n = 5 to 6). Data are expressed as mean ± SD; n = 5 to 6/group. Significance was determined by ANOVA followed by Tukey post hoc test (P < 0.05). *Significant difference between specimens from diabetic and matched NG animals; +significant difference between specimens from diabetic Col2α1Cre⁺.FOXO1^L/L and Col2α1Cre⁻.FOXO1^L/L mice.

Figure 5

Diabetes increases chondrocytes’ RANKL expression, which is FOXO1 dependent. Histologic sections were obtained 10 days after fracture from normoglycemic (NG) Col2α1Cre⁻.FOXO1^L/L mice (A), diabetic Col2α1Cre⁻.FOXO1^L/L mice (B), or diabetic Col2α1Cre⁺.FOXO1^L/L mice (C) with FOXO1 deletion. Immunofluorescence was carried out with antibody specific for RANKL, and no signal was detected with matched control antibody (not shown). Nuclei were stained with DAPI. The white dashed line marks the hypertrophic cartilage zone (HZ). Original magnification ×200. D: The percentage of RANKL-positive chondrocytes in the hypertrophic zone was counted in immunofluorescent images with antibody specific to RANKL from specimens obtained 10 and 16 days after fracture. E: RNA was obtained from fracture calluses 10 days after fracture, and RANKL mRNA levels were quantified by real-time PCR for NG Col2α1Cre⁻.FOXO1^L/L (black bar), diabetic Col2α1Cre⁻.FOXO1^L/L (white bar), and diabetic Col2α1Cre⁺.FOXO1^L/L (gray bar) mice (n = 5 to 6/group). D and E: *Significant difference between specimens from diabetic and NG Col2α1Cre⁻.FOXO1^L/L mice; +significant difference between specimens from diabetic Col2α1Cre⁺.FOXO1^L/L and Col2α1Cre⁻.FOXO1^L/L mice; #significant difference between the diabetic Col2α1Cre⁺.FOXO1^L/L and NG Col2α1Cre⁻.FOXO1^L/L mice. For each comparison significance was determined by ANOVA followed by Tukey post hoc test (P < 0.05). F: Primary chondrocytes were isolated from the control Col2α1Cre⁻.FOXO1^L/L and experimental Col2α1Cre⁺.FOXO1^L/L mice and grown to confluence. They were then incubated in standard media with serum for 14 h or cultured in chondrogenic media for 4 days. RNA was obtained, and RANKL mRNA levels were measured by real-time PCR. *Significant difference between cells obtained from Col2α1Cre⁺.FOXO1^L/L and Col2α1Cre⁻.FOXO1^L/L determined by Student t test (P < 0.05). G: Primary chondrocytes were isolated from Col2α1Cre⁻.FOXO1^L/L and Col2α1Cre⁺.FOXO1^L/L and incubated in standard media with low glucose (LG), HG, CML-BSA, advanced AGE, or unmodified BSA. RANKL was detected by immunofluorescence with antibody specific for RANKL, and no signal was detected with matched control antibody (not shown). Nuclei were stained with DAPI. * and + indicate a significant difference between the control and FOXO1-deleted chondrocytes under the same conditions by Student t test (P < 0.05).

Figure 6

FOXO1 nuclear localization, FOXO1 interaction with the RANKL promoter, and FOXO1-induced RANKL promoter activity in chondrocytes are stimulated by HG and an advanced AGE. A: Chondrocytes were incubated in standard media or media supplemented with HG (25 mmol/L) or osmotic control (OC), 25 mmol/L mannitol for 5 days, or 200 μg/mL CML-BSA, an AGE, or unmodified BSA for 3 days. ChIPs were performed by pulldown with FOXO1-specific antibody or matched control IgG. B: Chondrocytes were cotransfected with a FOXO1 expression plasmid and a RANKL luciferase reporter construct and incubated in standard media or media supplemented with HG or an AGE as above. Luciferase activity was measured and normalized by Renilla control. C: FOXO1 nuclear localization was measured by flow cytometry in chondrocytes cultured with standard glucose (LG) vs. HG for 5 days and compared with positive control, cells incubated in TNF-α (10 ng/mL) for 1 h. For the last 24 h, cells were incubated with the following inhibitors: 1L6-hydroxymethyl-chiro-inositol-2-(R)-2-O-methyl-3-O-octadecyl-sn-glycerocarbonate, an AKT inhibitor; TAK242, a TLR4 inhibitor; trichostatin A, a deacetylase inhibitor; NAC, an antioxidant; l-NAME, a nitric oxide synthase inhibitor; or insulin. D: FOXO1 nuclear localization was measured in chondrocytes stimulated with AGE, CML-BSA, or unmodified BSA for 3 days and compared with positive control, TNF-α. For the last 24 h, cells were incubated with inhibitors as described for C. E: FOXO1 nuclear translocation was measured by flow cytometry in chondrocytes incubated in HG for 5 days plus an AGE, CML-BSA, or unmodified BSA for 3 days and compared with cells incubated in TNF-α, a positive control. For the last 24 h, cells were incubated with inhibitors as described for C. Data are expressed as mean ± SEM. *Significant difference compared with standard media (LG) determined by ANOVA followed by Tukey post hoc test (P < 0.001); +significant difference compared with HG+AGE group determined by ANOVA followed by Tukey post hoc test (P < 0.001).