All animal studies were carried out with approval from the University of Pennsylvania.

Institutional Animal Care and Use Committee (IACUC) (Protocol # 803894) and the Guide for the care and use of laboratory animals, eighth edition (2011), were followed. FOXOL/L mice were provided by R.A. DePinho (MD Anderson Cancer Center, Houston, TX) and created as previously described (33). The experimental group included mice with lineage-specific FOXO1 deletion in chondrocytes (Col2α1Cre⁺.FOXO1^L/L) and results were compared with Col2α1Cre^-.FOXO1^L/L littermate controls. All animals were monitored daily by University Laboratory Animal Services (ULAR) and cages were changed weekly with 5001 Rodent Diet. (Purina Lab Diet, St. Louis, MO). Every cage contained two to five mice under standard conditions with 14-hours light/10-hours dark cycles. Prior to starting the experiment, genotyping was performed via PCR using both Cre and FOXO1 primers using genomic DNA extracted from the mice’s tails/ears. The results were also verified at the end of the experiment. Type-1 diabetes was induced as described by us (34) and developed originally by Like and Rossini (35). Intraperitoneal (IP) streptozotocin (STZ) injections (40 mg/kg; Sigma-Aldrich, St. Louis, Missouri, US) in 10mM citrate buffer were given once every day for five consecutive days. Vehicle alone was used for control mice. Ten days after the last injection, blood glucose levels were measured weekly from small lacerations in the mice tails from both groups. Mice with two consecutive blood glucose readings of >220 mg/dl were considered diabetic. The mice had been diabetic for at least three weeks prior to starting the experiment.

At 12-14 weeks old, a simple transverse fracture was induced. as previously described (36). Briefly, an incision was made in the knee and a 30-gauge spinal needle was inserted for fixation. A controlled, closed, simple, transverse fracture was created by blunt trauma to the middle of the femur at the mid-diaphyseal region. There was no change in the animal behavior noticed between the different groups. Animals were euthanized, and femurs were harvested 16 days after fracture. All the sites were evaluated radiographically and physically at euthanasia. Fractures that were not mid-diaphyseal or that were grossly comminuted were excluded from the experiment. Less than 5% of the samples were excluded. Fractured samples were fixed for 24 hours in cold 4% paraformaldehyde. Decalcification was achieved by incubating the samples at room temperature in 10% ethylenediaminetetraacetic acid (EDTA) solution for five weeks before embedding them in paraffin blocks. Transverse sections were prepared as described by us and initially by Gerstenfeld et al. (14, 36).

After dewaxing and hydration, a pressure cooker (2100-Retriever Aptum, Southampton, UK) was used at 120°C for 20 minutes with 10mM of citric acid with a pH of 6.0 for antigen retrieval followed by non-specific binding blocking for 55 minutes, using nonimmune serum matching the secondary antibody. Slides were incubated overnight with CD-31 specific antibody (Abcam, ab28364), anti-cleaved caspase 3 antibody (Cell Signaling Technology 96615) or the appropriate isotype-matched negative control IgG (Vector, I-1000, Burlingame, CA). The primary antibody was localized by a biotinylated secondary antibody (Vector, BA-1000). To localize the antibody complex, Alexa Fluor 546–conjugated streptavidin (Invitrogen S-11225, Carlsbad, CA) was used. Nuclei were counterstained with DAPI (Sigma-Aldrich, St. Louis, MO). Images were captured at different magnifications (40x, 200x, and 400x magnification) using a fluorescence microscope (ECLIPSE 90i; Nikon). The exposure time was set so that the IgG control had no signal. The quantification was performed with the aid of NIS Elements AR image analysis software. The unit of measure was the animal. Each value was calculated by examining six to eight animals in each group.

Apoptotic cells were detected by DeadEnd™ Colorimetric TUNEL System (Promega, WI, USA), which detects apoptotic cells by labeling and detecting DNA strand breaks by the TUNEL method. To distinguish apoptotic chondrocytes from other cells, the images were combined with a bright-field channel to confirm the cell morphology. In addition, the TUNEL-positive cells were compared with a safranin-o/fast green stain to verify the location of chondrocytes and define the entire region of interest. Slides were first deparaffinized and hydrated in the same manner mentioned. Slides were then incubated at room temperature for 15 minutes in diluted proteinase K solution and then rinsed with phosphate-buffered saline (PBS). This was followed by 5 minutes of incubation in endogenous oxidation-blocking solution, 3% hydrogen peroxide, at room temperature and then rinsed in PBS. After that, slides were incubated at room temperature for 30 seconds in equilibration buffer, then in working strength TdT enzyme for an hour at 37°C. Slides were then incubated in working strength stop/wash buffer for 10 minutes at room temperature, and they were then rinsed in PBS and incubated in anti-digoxigenin conjugate for 30 minutes at 37°C. Slides were rinsed again in PBS and mounted with DAPI to stain the nuclei. The mean number of immunopositive cells was calculated for each group examining six to eight animals per group. The number of immunopositive cells was divided by the area or as a percentage of the total number of cells in the region of interest.

ATDC5 chondrocytes obtained from the American Type Culture Collection (ATCC) (Manassas, VA, USA) were used for in vitro experiments. Cells were cultured in 50% Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Gaithersburg, MD, USA) and 50% F12 (Gibco) with 5% fetal bovine serum (FBS) and 1% Antibiotic-Antimycotic (Anti-Anti) (Thermo Fisher Scientific, Waltham, MA, US). Hypertrophic differentiation induction was performed using ascorbic acid (50mg/mL) for 6 days with increasing concentrations of NaH2PO4 0.5mM, 1mM, and 2mM (37). All cell cultures were maintained in a 5% CO2 humidified incubator at 37°C.

Quick-RNA MicroPrep kit (Zymo Research, Irvine, CA, USA) was used according to the manufacturer’s instructions to isolate RNA. RNA was converted to cDNA using an ABI High- Capacity RNA-to-cDNA kit (Applied Biosystems; cat# 4387406). caspase3 mRNA levels were measured by qPCR in ATDC5 chondrocytes cultured in low glucose (5mM d-glucose) and high glucose (25mM d-glucose) for 5 days and transfected with FOXO1 siRNA or scrambled siRNA. qPCR was performed using ABI Fast SYBR Green Master Mix (cat# 4385612) and a StepOne Plus real-time PCR system (Applied Biosystems). Relative amounts were calculated using the ΔΔCt method. Data are expressed as percent input after quantitative amplification of equivalent amounts of DNA. Experiments were performed with triplicate replicates and carried out three times with similar results.

siRNA transfections with ATDC5s were performed at approximately 60-70% confluence in 6-well plates with 10nM siFOXO1 or scramble control (Dharmacon, Lafayette, CO) using GenMute transfection reagent (Rockville, MD) according to the manufacturer’s instructions. Plasmid transfections were performed in OptiMEM (Gibco) medium with 1250ng plasmid in 3.75ul of Lipofectamine 3000 transfection reagent per well and following manufacturer’s instructions with ATDC5s at approximately 60-70% confluence for 4.5 hours before being replaced with full media. To quantify caspase-3 expression, cells were co-transfected with treatment vectors: empty, ADA [Threonine 24 to Alanine (A) and Serine 253 to Aspartate (D) and Serine 316 to Alanine (A)], ADA + 6KQ (K242, K245, K259, K262, K271, and K291 were replaced with glutamine on ADA), ADA + 6KR (K242, K245, K259, K262, K271, and K291 were replaced with arginine on ADA), KQ(K242, K245, K259, K262, K271, and K291 were replaced with glutamine) or KR(K242, K245, K259, K262, K271, and K291 were replaced with arginine) mutants (Addgene, Cambridge, MA. pCMV5 backbone) along with a caspase-3 luciferase reporter (pGL4) as described and generously provided by Dr. Estrov (38). Co-transfection utilized the same transfection protocol with an expression vector, and the reporter was added at a 1:1 ratio of 150ng per well in a 48-well plate. Expression values were normalized using a Renilla control (pGL4) containing the CMV promoter at a 1:20 ratio. Results were quantified using a Dual-Luciferase Reporter Assay Kit from Promega (Madison, WI, cat# - E1960) and quantified on a Tecan Infinite M200. Experiments were performed with triplicate replicates and carried out three times with similar results.

Chromatin immunoprecipitation (ChIP) assays were performed with the ChIP-IT Kit (Active Motif, Carlsbad, CA) using approximately 1.5 x107 ATDC5 cells. The cells were cultured using multiple conditions, including 1) hypertrophic differentiation for 6 days using differentiation media as mentioned earlier; 2) cells at 70–80% treated with CML-BSA (200 mg/mL), an AGE, for 3 days or unmodified BSA (200 mg/mL) for a similar period; and 3) cells grown in High glucose (HG) (25 mmol/L) media for 5 days. Formaldehyde was used to fix the cells and nuclei obtained following Dounce homogenization. ChIP was performed following the manufacturer’s instructions using an anti-FOXO1 antibody (5 mg) (SC-11350X; Santa Cruz Biotechnology) or control polyclonal non-specific IgG (Cell Signaling Technology). Protein G-coupled beads were used to purify the chromatin–antibody complexes. Three quantitative real-time PCR reactions for the caspase-3 promoter region, which contains FOXO1 consensus response elements, were done with similar results. Experiments were performed with triplicate replicates and carried out three times with similar results.

We have previously examined the effect of diabetes on angiogenesis during fracture healing by identifying blood vessels with an antibody to Factor VIII or CD31 (39). As shown in Table 1 , which was adapted from (39), there was a 65%–80% increase in the number of blood vessels between day 10 and day 16 in normoglycemic mice (p < 0.05) ( Table 1 ). Diabetes reduced the number of Factor VIII⁺ small and moderate-sized blood vessels by almost half (p >0.05) ( Table 1 ). Mechanistically, this was related to the level of inflammation as shown by the rescue of diabetes-reduced neovascularization when a TNFα-specific inhibitor, pegsunercept (PEG), was applied.

Since we had shown that some of the effects of TNF could be directly related to the transcription factor FOXO1 (40), we examined the hypothesis that FOXO1 is a key factor in regulating angiogenesis in diabetic animals. This was accomplished by studying neovascularization in healing fractures in animals with chondrocyte-specific FOXO1 deletion in diabetic experimental (Col2a1Cre⁺.FOXO1^L/L) compared to diabetic control (Col2a1Cre^–.FOXO1^L/L) littermates. Diabetes resulted in a ~ 50% reduction in the number of both small and moderately sized vessels compared to the WT ( Figure 1 , P > 0.05). Conditional FOXO1 deletion in chondrocytes in diabetic animals resulted in a partial and significant rescue of diabetes-inhibited formation of small and moderate-sized blood vessels (P<0.05).

We have previously examined apoptotic cells in fracture healing using a TUNEL assay (40). Table 2 , adapted from (40), demonstrated that the number of apoptotic chondrocytes was 5.4-fold higher in the diabetic group compared to the normoglycemic (P<0.05; Table 2 ). Insulin treatment significantly reduced most of this increase (P<0.05). When the entire callus was examined, the diabetic group showed a 2.5-fold increase in the total number of apoptotic cells compared to the normoglycemic, which was largely rescued by insulin treatment (P<0.05; Table 2 ). The increase in apoptosis was directly linked to the effect of TNF-α as demonstrated by a complete rescue when TNF-α was inhibited by pegsunercept. There was also no significant difference in the number of apoptotic chondrocytes between the diabetic and the normoglycemic group on day 10. Chondrocyte apoptosis increased further on day 16 post-fracture in the diabetic animals, and the increase was rescued to normal levels upon PEG treatment (P<0.05; Table 2 .)

We then examined whether the transcription of FOXO1 mediated apoptosis in diabetic fracture healing using the experimental animals described above. There was a ~ 50% increase in TUNEL-positive chondrocytes in the hypertrophic and the mixed zone that contains both cartilage and bone ( Figures 2A–E , p<0.05). This increase was completely rescued to normal levels upon specific FOXO1 deletion in chondrocytes in diabetic animals ( Figure 2F , p<0.05).

To further examine how diabetes could increase chondrocyte apoptosis, we measured the levels of cleaved caspase-3. In vivo immunofluorescence with an antibody specific for cleaved caspase-3 showed that diabetes increased caspase-3 positive chondrocytes by ~50% compared to fracture healing in normoglycemic animals ( Figure 2D , p<0.05). FOXO1 deletion in chondrocytes in diabetic animals reduced the over-expression of caspase-3 to levels similar to that of the normal group ( Figure 2D , P>0.05).

We then directly assessed the role of FOXO1 in regulating caspase-3 expression by transfecting chondrocytes with FOXO1 siRNA or scrambled siRNA in low-glucose (LG) and high-glucose-containing media (HG). HG media resulted in a 2.7-fold increase in caspase-3 expression ( Figure 3A , p<0.05). Silencing FOXO1 with siRNA significantly downregulated the increase in caspase-3 expression induced by HG. ChIP assays were undertaken to measure FOXO1 interaction with the caspase-3 promoter. HG stimulated an ~ 8-fold increase in FOXO1 binding to the caspase-3 promoter ( Figure 3B , p<0.05). Treatment of chondrocytes with an advanced glycation end product (AGE), carboxymethyl lysine modified BSA, stimulated a similar increase. Luciferase reporter assays were undertaken to examine the direct effect of HG and AGE on caspase-3 promoter activity. Transfection of chondrocytes with a FOXO1 expression vector increased caspase-3 transcriptional activity 4-fold in the presence of HG compared to standard (LG) media and 6-fold in the presence of AGE compared to unmodified BSA ( Figure 3C , p<0.05). Interestingly, the combination of HG or AGE with FOXO1 transfection was greater than the effect of FOXO1 transfection alone.