Chondrocytes Promote Vascularization in Fracture Healing Through a FOXO1-Dependent Mechanism

Abstract

Chondrocytes play an essential role in fracture healing by producing cartilage, which forms an anlage for endochondral ossification that stabilizes the healing fracture callus. More recently it has been appreciated that chondrocytes have the capacity to produce factors that may affect the healing process. We examined the role of chondrocytes in angiogenesis during fracture healing and the role of the transcription factor forkhead box-O 1 (FOXO1), which upregulates wound healing in soft tissue. Closed fractures were induced in experimental mice with lineage-specific FOXO1 deletion by Cre recombinase under the control of a collagen-2α1 promoter element (Col2α1Cre⁺FOXO1^L/L) and Cre recombinase negative control littermates containing flanking loxP sites (Col2α1Cre^–FOXO1^L/L). Experimental mice had significantly reduced CD31⁺ new vessel formation. Deletion of FOXO1 in chondrocytes in vivo suppressed the expression of vascular endothelial growth factor-A (VEGFA) at both the protein and mRNA levels. Overexpression of FOXO1 in chondrocytes in vitro increased VEGFA mRNA levels and VEGFA transcriptional activity whereas silencing FOXO1 reduced it. Moreover, FOXO1 interacted directly with the VEGFA promoter and a deacetylated FOXO1 mutant enhanced VEGFA expression whereas an acetylated FOXO1 mutant did not. Lastly, FOXO1 knockdown by siRNA significantly reduced the capacity of chondrocytes to stimulate microvascular endothelial cell tube formation in vitro. The results indicate that chondrocytes play a key role in angiogenesis which is FOXO1 dependent and that FOXO1 in chondrocytes regulates a potent angiogenic factor, VEGFA. These studies provide new insight into fracture healing given the important role of vessel formation in the fracture repair process.

Introduction

Fractures of the long bones heal by intramembranous bone formation and by endochondral ossification that leads to formation of a stabilizing callus around the bone.⁽¹⁾ The callus is derived from periosteal mesenchymal stem cells (MSCs) that differentiate into chondrocytes, osteoprogenitors, and osteoblasts.^(2–4) Chondrocytes produce cartilage matrix, which forms an anlage for subsequent bone formation that is critical in stabilizing the fracture site. In addition to producing cartilage, chondrocytes may also affect healing by producing biologic factors. Chondrocytes have been shown to produce factors such as osteoprotegerin and RANKL that modulate cellular processes in bone remodeling.⁽¹⁾ However, the role of chondrocyte-produced factors in fracture healing has received relatively little attention.

Vascularization is critical for normal fracture healing,^(5–7) as shown by impaired healing if vascularization is inhibited,⁽⁸⁾ and improved healing when angiogenesis is enhanced.⁽⁹⁾ Angiogenesis occurs when endothelial cells migrate from existing vessels, proliferate, and subsequently form vessels.⁽¹⁰⁾ In the callus, new vessel formation provides a nutrient supply for cells, facilitates removal of cartilage by the delivery of osteoclast precursors, and contributes to healing by producing mediators.⁽¹¹⁾ Vascular cells produce a number of paracrine factors that impact chondrocytes and osteoblasts and affect bone production and the formation and maintenance of bone marrow.^(12–14) An important angiogenic factor is vascular endothelial growth factor-A (VEGFA), which is produced by mesenchymal stem cells, vascular cells, chondrocytes and osteoblasts.^(1,15,16)

FOXO1 is a forkhead box-O (FOXO) transcription factor that is expressed in chondrocytes and has been shown to modulate downstream gene targets in chondrocytes or chondrocyte activity.^(17–20) There are three principal FOXO family members. FOXO1 and FOXO3 are found in both bone and cartilage, whereas FOXO4 is considerably less abundant in these tissues.^(21,22) In chondrocytes FOXO1 has been shown to have a more dramatic effect than FOXO3.⁽²¹⁾ FOXO1 typically regulates gene expression by binding to DNA and modulating transcription.⁽²³⁾ The impact of FOXO1 is difficult to predict because its downstream targets are modified by the microenvironment.⁽²⁴⁾ In osteoblasts FOXO1 can be protective under aging conditions by inducing anti-oxidative factors,^(25,26) whereas under homeostatic conditions FOXO1 can modulate proliferation and differentiation.⁽²⁷⁾ In conditions of high inflammation FOXO1 can promote cell death,⁽²⁸⁾ whereas in normal conditions FOXO1 protects chondrocytes.^(17,18) Deletion of multiple FOXOs (FOXO1, FOXO3, and FOXO4) results in abnormal growth plate organization and skeletal abnormalities.⁽²⁹⁾ In soft tissue healing deletion of FOXO1 in keratinocytes of normal wounds impairs re-epithelialization whereas deletion of FOXO1 in keratinocytes of diabetic wounds accelerates it.^(30,31) This result suggests that in normal conditions FOXO1 acts to promote soft tissue wound healing. Fracture healing stimulates nuclear localization of FOXO1 in chondrocytes, also suggesting that FOXO1 in chondrocytes may affect the fracture healing process.^(19,20)

We sought to determine if FOXO1 in chondrocytes affected angiogenesis during fracture healing given the importance of neovascularization on endochondral ossification.^(6,15) Given the complex intercellular and microenvironment interactions that occur during fracture healing, we induced closed fracture of the femur in mice with a lineage-specific FOXO1 deletion in chondrocytes driven by the collagen-2α1 promoter (Col2α1-Cre⁺FOXO1^L/L) and control mice still expressing FOXO1 in chondrocytes (Col2α1Cre⁻FOXO1^L/L) in order to better mirror human biology. Results show the FOXO1 ablation in chondrocytes leads to significantly reduced VEGFA expression and angiogenesis. FOXO1 binds to the VEGFA promoter in chondrocytes and FOXO1 induces VEGFA transcriptional activity. Conversely, knockdown of FOXO1 reduces VEGFA expression by chondrocytes in vitro and significantly impairs the capacity of chondrocytes to support vessel formation in microvascular endothelial cells. Thus, activation of FOXO1 in chondrocytes plays an important role in angiogenesis in fracture healing.

Subjects and Methods

Animals and fracture induction

All animal studies were carried out with approval from the University of Pennsylvania Institutional Animal Care and Use Committee (IACUC). 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, USA). Experiments were performed with littermate controls, and mice were housed with up to five littermates of the same sex. Mice were not kept singly housed. FOXO^L/L mice were provided by RA DePinho (MD Anderson Cancer Center, Houston, TX, USA) and created as described.⁽³²⁾ Col2α1Cre⁺ mice were provided by Patrick O’Connor (Rutgers New Jersey Medical School, Newark, NJ, USA) and created as described.⁽³³⁾ Col2α1Cre⁺ were bred with FOXO^L/L to generate the experimental mice (Col2α1Cre⁺FOXO1^L/L) and littermate (Col2α1Cre⁻FOXO1^L/L) controls. Genotypes were determined by PCR using primers specific for Cre-Recombinase (5′-ATCCAGGTTACGGATATAGT-3′ and 5′-ATCCGAAAAGAAAACGTTGA-3′) and specific for FOXO1 (5′-GCTTAGAGCAGAGATGTTCTCACATT-3′, 5′-CCAGAGTCTTTGTA TCAGGCAAATAA-3′, and 5′-CAAGTCCATTAATTCAGCACATTGA-3′). Experimenters were not blinded when placing mice into experimental groups but were blinded during fracture performance and endpoint harvesting. The experimental mice had no obvious skeletal abnormalities and had long bones that were indistinguishable from controls. A simple transverse fracture was performed on the femur of 12-week-old adult male mice as described⁽³⁴⁾ and a 27G spinal needle was inserted and used to support the femur during healing. All fractures that were comminuted and not in the mid-diaphyseal region were excluded from the study. Animals were euthanized 16 days postfracture, and the fracture callus was harvested, after removal of tissues and soft musculature surrounding the callus. To evaluate the fracture healing, overall callus volume was determined by micro–CT (μCT). For μCT analysis there was a minimum of nine animals examined per group with specimens obtained 35 days after fracture with a Scanco μCT40 (Scanco Medical AG, Brüttisellen, Switzerland) at a voxel size of 19 μm. Scan settings were 70 kVp, 114 mA, and 200 ms integration time.

In vivo sample preparation

Fracture samples were fixed for 24 hours in cold 4% paraformaldehyde, decalcified for 5 weeks in 10% EDTA, processed, and embedded in paraffin blocks. Transverse paraffin-embedded sections were prepared as described.⁽³⁵⁾ To obtain RNA, 16 days postfracture six WT and six KO calluses were snap frozen in liquid nitrogen, ground into powder in the presence of liquid nitrogen using a sterilized mortar and pestle, and isolated with a kit from Ambion (distributed by Thermo Fisher Scientific, Waltham, MA, USA; Cat# AM1912) per the manufacturer’s instructions. RNA yield was determined on Tecan Infinite M200 (Tecan, Männedorf, Switzerland). RNA was converted to cDNA using Applied Biosystems (ABI) High-Capacity RNA-to-cDNA kit (Applied Biosystems, Foster City, CA, USA; cat# 4387406) and qPCR was performed using ABI Fast SYBR Green Master Mix (Applied Biosystems; cat# 4385612) on StepOne Plus real-time PCR system (Applied Biosystems). Analysis was performed using the delta delta threshold cycle (ΔΔCT) method, and statistical tests were run on Prism (GraphPad Software, Inc., La Jolla, CA, USA).

Immunofluorescence

Antigen retrieval was achieved at 120°C (2100-Retriever; Aptum, Southampton, UK) in 10mM, pH 6.0 citrate buffer. Tissue sections were incubated with VEGFA-specific antibody (Abcam, Cambridge, MA, USA; ab46154) or CD-31 specific antibody (Abcam; ab28364) overnight at 4°C or with matched control IgG (Vector, Burlingame, CA, USA; I-1000). Primary antibody was localized by a biotinylated secondary antibody (Vector; BA-1000). Sections were subsequently incubated with avidin-biotin peroxidase enzyme complex (ABC; Vector Laboratories) and followed by tyramide signal amplification (TSA; PerkinElmer, Waltham, MA, USA) to amplify signal and Alexa Fluor 546–conjugated streptavidin (Invitrogen, Carlsbad, CA, USA; S-11225). Nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI; Sigma-Aldrich, St. Louis, MO, USA). Regions of interest were defined by Safranin-O/fast green histostains to establish areas of cartilage and bone and to identify the transition zone. For immunofluorescence the exposure time was set so that the IgG-negative control images were negative. Immunofluorescent images were captured at magnification indicated by fluorescence microscopy (ECLIPSE 90i; Nikon, Tokyo, Japan). The percent of immunopositive cells was determined based on the total number of cells established by DAPI counterstain.

Small blood vessels were identified by CD31 immunofluorescence as two to four CD31⁺ cells per vessel and moderate vessels were counted as having five to eight CD31⁺ cells per vessel. The regions of interest were the cartilage for VEGFA immunofluorescence and the transition zone consisting of new endochondral bone formation adjacent to cartilage to assess CD31 immunopositive blood vessels. These areas were identified with assistance from in adjacent sections stained with safranin-O/fast green.

Cell culture

In vitro experiments were performed with ATDC5 chondrocytes from American Type Culture Collection (ATCC) (Manassas, VA, USA). Cells were cultured in 50% DMEM (Gibco, Gaithersburg, MD, USA) and 50% F12 (Gibco) with 5% FBS. Differentiation to chondrocytes was performed over 6 days with ascorbic acid (50 μg/mL) and increasing concentrations of NaH₂PO₄ 0.5mM, 1mM, and 2mM as described.⁽³⁶⁾ Human microvascular endothelial cells (HMVECs) were purchased from Cell Systems (Kirkland, WA, USA) and cultured in EGM2-MV medium (Lonza, Walkersville, MD, USA) with 5% FBS and 1% antibiotics in an incubator with 5% CO₂ at 37°C.

Transfection and dual luciferase assay

siRNA transfections were performed with cells at approximately 60% to 70% confluence in six-well plates with 10nM siFOXO1 or scramble control (Dharmacon, Lafayette, CO, USA) using GenMute transfection reagent (Rockville, MD, USA) following the manufacturer’s instructions. All plasmid transfections were performed in OptiMEM (Gibco) medium with 1250 ng plasmid in 3.75 μL of Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific) per well following the manufacturer’s instructions for 4.5 hours before being replaced with full media. To quantify VEGFA expression cells were co-transfected with one of the following: empty vector alone; FOXO1-ADA (Threonine 24 to Alanine [A] and Serine 253 to Aspartate [D] and Serine 316 to Alanine [A]); FOXO1-ADA-KQ (K242, K245, K259, K262, K271, and K291 were replaced with glutamine on ADA); or FOXO1-ADA-KR (K242, K245, K259, K262, K271 and K291 were replaced with arginine on ADA) mutants (Addgene, Cambridge, MA, USA. Cat# 12149, 17563, and 17564, respectively, pCMV5 backbone) along with a VEGFA luciferase reporter (pGL3) containing 2324 bp of the VEGFA reporter as described and generously provided by Dr. Fraizer.⁽³⁷⁾ Co-transfection utilized the same transfection protocol with expression vector and reporter added at a 1:1 ratio (150 ng per well in a 48-well plate). Expression values were normalized using a Renilla control (pGL3) containing the CMV promoter at a 1:20 ratio. Results were quantified using Dual Luciferase Reporter Assay Kit from Promega (Madison, WI, USA, cat# E1960) and quantified on a Tecan Inifite M200.

Western blots

Western blots were performed as described.⁽³⁸⁾ Whole-cell lysate from differentiated ATDC5 cells were isolated using a kit from Active Motif (Carlsbad, CA, USA; cat# 40010) per the manufacturer’s instructions. Antibodies used were purchased from Cell Signaling Technology (FOXO1 C29H4 Rabbit mAB; Danvers, MA, USA; cat# 2880s) at 1 μg/mL, B-actin (Sigma Aldrich; cat# A5441) at 1 μg/mL, and VEGFA antibody (Abcam; cat# ab46154) at 1 μg/mL. Secondary anti-rabbit IgG–horseradish peroxidase (IgG-HRP) antibody was purchased from GE Healthcare Life Sciences (Marlboro, MA, USA; cat# NA934) and used at 0.33 μg/mL. To identify anti-B-actin primary mouse monoclonal antibody, an anti-mouse IgG-HRP was purchased from Cell Signaling Technology (cat# 7076S) and used at 0.33 μg/mL. Images were detected using Pierce ECL Western Blotting Substrate (Pierce, Rockford, IL, USA; Cat# 32209) and subsequently analyzed with a GE iQuant 4000 imager (GE Healthcare, Piscataway, NJ, USA). FOXO1 and VEGFA had apparent molecular sizes of 78 kD and 42 kD, respectively. All analysis was performed with image analysis and background subtraction.

RNA isolation, cDNA conversion, chromatin immunoprecipitation assay, qPCR

RNA isolation was performed using Quick-RNA MicroPrep kit (Zymo Research, Irvine, CA, USA) according to the manufacturer’s instructions. RNA was converted to cDNA using ABI High-Capacity RNA-to-cDNA kit (Applied Biosystems; cat# 4387406). Chromatin immunoprecipitation (ChIP) assays were carried out with a Chromatin IP DNA Purification Kit from Active Motif (cat# 58002) according to the manufacturer’s instructions with FOXO1 (FKHR) antibody purchased from Santa Cruz Biotechnology (Dallas, TX, USA; SC-11350) used for pull-down. Endpoint analysis examined the VEGFA promoter region (–735 bp to –935 bp relative to the initiation start site) which contains FOXO1 consensus response elements using primers: forward: 3′-GGAGAATGGGAGGACAGAGTA-5′; reverse: 3′-GTAGAGGCT CCCTCAGTTCTA-5′. To determine the role of acetylation on VEGFA, ATDC5 chondrocytes were incubated with 100nM Trichostatin A (Cayman Chemical, Ann Arbor, MI, USA; cat #89730) to inhibit histone deacetylase activity (HDAC) and 1 μM PF-CBP1 (Cayman Chemical; cat #18811) to inhibit p300 activity for 24 hours before RNA isolation. VEGFA mRNA levels were measured by qPCR in ATDC5 chondrocytes transfected with FOXO1 expression vectors, FOXO1-ADA, FOXO1-ADA-KQ, or FOXO1-ADA-KR 48 hours before RNA isolation, as described in Transfection and Dual Luciferase Assay. 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.

Tube formation assay

Chondrocyte conditioned media was collected after 48 hours of incubation with transfected ATDC5 cells. Reduced Growth Factor Matrigel (Corning, Corning, NY, USA) was used to coat 96-well plates according to the manufacturer’s instructions (50 μL at 10 mg/mL for 45 min, 5% CO₂ at 37°C. HMVECs were grown to 80% confluence in EGM2-MV media with full supplementation (Lonza) and then incubated in endothelial basal medium (EBM) (Lonza) with 0.1% FBS for 24 hours. HVMECs were added to Matrigel at 2 × 10⁴ per well as described.⁽³⁹⁾ HMVECs were then incubated in ATDC5 chondrocyte conditioned media without or with VEGF at 150 pg/mL (R&D Systems, Minneapolis, MN, USA; Cat# 293-VE-010). The number of tubes per well were counted by a blinded examiner with NIS Elements software (Nikon) after 18 hours.

Statistical analysis

For in vitro and in vivo assays with multiple comparisons ANOVA with Tukey’s post hoc test was carried out. For comparisons between two groups Student’s t test for unpaired samples was performed. For all tests significance was set at p < 0.05. For in vitro assays a minimum of three samples were examined per group and three independent experiments were carried out with similar results. For in vivo immunofluorescence experiments there were a minimum of six specimens per group.

Results

FOXO1 deletion in chondrocytes in vivo results in reduced VEGFA⁺ hypertrophic chondrocytes, CD31⁺ cells, and reduced VEGFA in the total fracture cell population.

We examined the impact of chondrocytes in new vessel formation during fracture healing in experimental Col2α1Cre⁺. FOXO1^L/L (Cre⁺) mice with FOXO1 deletion and control Col2α1Cre⁺. FOXO1^L/L (Cre⁻) littermates without FOXO1 deletion. The Col2α1Cre transgene has been shown to specifically delete several floxed genes in chondrocytes but no other cell types.^(40–44) Col2α1-driven Cre recombinase deleted FOXO1 determined by genotyping (Fig. 1A) and deleted FOXO1 in chondrocytes as determined by immunofluorescence with specific antibody to FOXO1 (Fig. 1B). μCT was performed 35 days after fracture to evaluate the fracture healing. FOXO1 deletion in chondrocytes resulted in a decrease in total callus volume (Fig. 1C). FOXO1 deletion in chondrocytes substantially reduced chondrocyte expression of VEGFA (Fig. 2A, B). In the WT mice 46% of the total population of chondrocytes were VEGFA-positive, but in FOXO1 ablated mice less than 3% of the chondrocytes expressed VEGFA (Fig. 2C, p < 0.05). Chondrocytes were an important source of VEGFA because VEGFA mRNA levels were significantly reduced in total RNA prepared from the whole callus of experimental mice with chondrocyte-specific FOXO1 deletion compared to matched littermate controls (Fig. 2D, p < 0.05).

Fig. 1.

Col2α1-driven Cre recombinase deletes FOXO1 in chondrocytes. (A) FOXO1 with flanking loxP sites (149 bp) is detected in both the Cre⁺ and matched Cre⁻ control mice whereas cleaved FOXO1 (198 bp) is detected only in mice with Cre recombinase. The Cre recombinase amplicon (352 bp) is detected in the Cre⁺ but not Cre⁻ mouse. (B) Immunofluorescent images with FOXO1 antibody or with matched control IgG. Safranin-O/fast green–stained histologic images show the corresponding region of interest. Original magnification ×100. (C) 35 days after fracture, callus volume was determined by μCT analysis (n ≥ 9 per group). Results are expressed as the mean ± SE. *p < 0.05 compared to the matched control group. Cont = control.

Fig. 2.

FOXO1 deletion in chondrocytes in vivo reduces VEGFA expression. Histologic sections were prepared from specimens obtained 16 days postfracture. (A, B) Immunofluorescent images with VEGFA antibody or with matched control IgG. Safranin-O/fast green–stained histologic images show the corresponding region of interest. Original magnification, ×10 (A) or ×40 (B). (C) Quantitation of VEGFA immunopositive chondrocytes. n = 6 animals per group. (D) RNA was isolated and measured by qPCR. n = 6 animals per group. Results are expressed as the mean ± SE. *p < 0.05 compared to the matched control group. Cont = control.

To examine the impact of chondrocyte-specific deletion of FOXO1 on angiogenesis, the number of CD31⁺ vessels was measured in the transition zone adjacent to cartilage. Conditional FOXO1 deletion reduced the number of vessels (Fig. 3A). Quantitation showed a ~70% decrease in the number of moderate and small vessels compared to the WT mice, showing a significant effect (Fig. 3B, C; p < 0.05).

Fig. 3.

Deletion of FOXO1 in chondrocytes reduces angiogenesis during fracture healing. Angiogenesis was measured by immunofluorescence with antibody specific for CD31⁺ compared to control IgG in vivo 16 days postfracture. (A) Immunofluorescent images with CD31⁺ antibody or with matched control IgG. Safranin-O/fast green–stained histologic images show the corresponding region of interest. Original magnification, ×20. (B, C) Quantitation of CD31⁺ blood vessels in the transition zone adjacent to cartilage. Small blood vessels (two to three CD31⁺ cells/vessel) and moderate-sized vessels (five to eight CD31^þ cells/vessel) were counted. n = 6 animals per group. Results are expressed as the mean ± SE. *p < 0.05 compared to wild type (WT) group.

FOXO1 binds to the VEGFA promoter

To establish whether FOXO1 interacts directly with the VEGFA promoter a ChIP assay was performed (Fig. 4A). Pull-down with a FOXO1 antibody showed fivefold enrichment of an amplicon that included the FOXO1 consensus response element compared to the pull-down with IgG control. Chondrocytes were incubated with Trichostatin A (HDAC inhibitor) or PF-CBP1 (p300 inhibitor) to determine the impact of inhibiting deacetylation and acetylation, respectively. In the presence of a deacetylase inhibitor FOXO1 binding to the VEGFA promoter decreased dramatically and was not significantly different than the IgG control. In contrast, incubation of cells with a p300 acetylation inhibitor increased FOXO1 interaction with the VEGFA promoter 6.4-fold (p < 0.05), which was significantly higher than FOXO1 binding to the promoter without inhibitor (Fig. 4A).

Fig. 4.

FOXO1 directly binds to the VEGFA promoter, which is modulated by acetylation status. (A) Chromatin immunoprecipitation assay was performed with ATDC5 chondrocytes with pull-down by FOXO1-specific antibody or control IgG and PCR amplification was performed for a region of the VEGFA promoter flanking the FOXO1 consensus response element. (B) ATDC5 chondrocytes were co-transfected with empty vector or pcFOXO1; FOXO1 constructs ADA, ADA+6KQ, or ADA+6KR; and a VEGFA luciferase reporter and Renilla control construct. (C) ATDC5 chondrocytes were transfected with empty vector (pCMV5) or pcFOXO1 (WT FOXO1), FOXO1constructs ADA (constitutively active), ADA+6KQ (constitutively active, acetylated mutant), or ADA+6KR (constitutively active, deacetylated mutant). VEGFA mRNA levels were measured by qPCR. (D) ATDC5 chondrocytes were treated with HDAC inhibitor (TSA) and p300 inhibitor (CBP). VEGFA mRNA levels were measured by qPCR. Results are expressed as the mean ± SE. #p < 0.05 compared to control IgG; ##p < 0.05 compared to control FOXO1; *p < 0.05 compared to control group; *p < 0.05 compared to matched FOXO1 non-mutant.

FOXO1 stimulates transcriptional activity of VEGFA

To investigate FOXO1 regulation of VEGFA promoter activity, cotransfection experiments were carried out and analyzed via qPCR and luciferase reporter assays. To examine mRNA levels of VEGFA, chondrocytes were transfected with a modified FOXO1 (ADA) that is constitutively active or with matched mutated FOXO1 expression vectors representing deacetylated (ADA+KR) or acetylated (ADA+KQ) FOXO1 isoforms. VEGFA promoter activity was examined by luciferase reporter assay (Fig. 4B). Chondrocytes were co-transfected with a VEGFA luciferase reporter construct and a WT FOXO1, FOXO1-ADA, FOXO1-ADAKR, or FOXO1-ADA-KQ expression vector. Overexpression of WT FOXO1 in chondrocytes increased VEGFA promoter activity fourfold (Fig. 4B; p < 0.05). Transfection with a constitutively active FOXO1 increased VEGFA reporter activity 10-fold (p < 0.05). In contrast, an acetylated isoform of FOXO1 induced only a small 1.5-fold increase (p < 0.05) in VEGFA transcriptional activity whereas the deacetylated isoform of FOXO1 stimulated a 10-fold increase, similar to the constitutively active FOXO1 (Fig. 4B). The effect of transfection on mRNA levels of VEGFA were also measured (Fig. 4C). Exogenous WT FOXO1 induced a 1.2-fold increase in VEGFA expression (Fig. 4C; p < 0.05). FOXO1-ADA increased VEGFA 1.7-fold compared with control (Fig. 4C; p < 0.05). ADA+KR did not increase VEGFA expression (Fig. 4C; p > 0.1) whereas FOXO1-ADA-KQ increased VEGFA 1.6-fold (Fig. 4C; p < 0.05). These results are consistent with luciferase reporter assays, although the magnitude is less, most likely due to the impact of transfection efficiency. The impact of acetylation and deacetylation was also examined using specific inhibitors. Result show that Trichostatin A (HDAC inhibitor) reduced the induction of VEGFA mRNA levels by 60% (p < 0.05) and that PF-CBP1 (P300 inhibitor) stimulated a 70% increase (Fig. 4D; p < 0.05), consistent with overexpression results obtained with FOXO1 mutants and results obtained with ChIP assays.

FOXO1 knockdown regulates chondrocyte-induced microvascular endothelial tube formation

Knockdown of FOXO1 by siRNA reduced VEGFA mRNA levels by 55% (Fig. 5A; p < 0.05). Using siRNA, FOXO1 knockdown also reduced VEGFA at the protein level as determined by Western blot analysis (Fig. 5B). The conditioned medium from ATDC5 chondrocytes was tested for induction of microvascular endothelial cell tube formation. Scrambled siRNA had no effect on the capacity of chondrocytes to stimulate tube formation (Fig. 5C, p < 0.05). In contrast, FOXO1 siRNA reduced tube formation by 62% (p < 0.05). Adding exogenous VEGFA to the media conditioned by chondrocytes with FOXO1 knockdown increased tube formation twofold (p < 0.05).

Fig. 5.

Chondrocytes express VEGFA and stimulate tube formation regulated by FOXO1. (A) ATDC5 chondrocytes were transfected with FOXO1 siRNA or scrambled siRNA and VEGFA mRNA was measured 48 hours later by qPCR. (B) ATDC5 chondrocytes were transfected with siFOXO1 or siSCR and VEGFA was measured by Western blot. All images were from same blot: FOXO1 (78 kD), VEGFA (42 kD), and B-actin loading control (37 kD). (C) The tube assay was performed with human microvascular endothelial cells incubated in Matrigel with conditioned media without or with exogenous VEGFA from chondrocytes transfected with siRNA as indicated. Results are expressed as the mean ± SE. *p < 0.05 compared with siSCR control group; NT = non-transfected; siFOXO1 = FOXO1 siRNA; siSCR = scrambled siRNA.

Discussion

To investigate how chondrocytes may contribute to vascularization we examined mice with lineage-specific deletion of FOXO1 in chondrocytes. These studies revealed that chondrocytes play an important role in vascularization of the fracture callus and that the transcription factor FOXO1 regulates this effect. Effective Cre recombinase deletion of FOXO1 was shown by PCR in experimental animals by genotyping and in chondrocytes specifically by immunofluorescence. Our data shows that through FOXO1, chondrocytes in fracture healing play a significant role in the healing response by induction of VEGFA. This is likely because FOXO1 deletion in chondrocytes significantly reduced VEGFA expression in these cells and reduced vascularization in the transition zone adjacent to cartilage. Chondrocytes in vitro were also shown to express VEGFA in a FOXO1-dependent manner. Moreover, ChIP assays established that FOXO1 binds to the VEGFA promoter and co-transfection experiments indicate that FOXO1 regulates VEGFA transcriptional activity.

Chondrocyte-specific deletion of FOXO1 decreased the number of CD31-positive small and moderate-sized blood vessels examined in histologic sections. CD31 is expressed in a range of endothelial cells and its expression in blood vessels accurately reflects the level of angiogenesis.⁽⁴⁵⁾ Ablation of FOXO1 in chondrocytes reduced overall VEGFA mRNA levels in the fracture callus, indicating that chondrocytes are an important source of VEGFA during fracture healing. When VEGFA was examined in histologic sections by immunofluorescence, the number of chondrocytes with detectable VEGFA expression was substantially reduced in fractures of experimental mice compared to littermate controls. These studies show that chondrocytes are an important cell source of VEGFA and add to previous reports that osteoblasts express VEGF.⁽¹⁴⁾

Tube formation occurs during the later stages of angiogenesis and is a necessary step in blood vessel formation.^(46,47) In vitro assays show that chondrocytes stimulate tube formation, which was significantly reduced with FOXO1 knockdown. The results indicate that FOXO1 is necessary for the production of proangiogenic factors by chondrocytes, which induce tube formation.⁽⁴⁸⁾ Moreover, rescue experiments suggest that VEGFA is an important factor in FOXO1-driven angiogenesis in chondrocytes. Thus, chondrocytes during fracture healing produce mediators that promote angiogenesis, consistent with reports that they induce angiogenesis in osteoarthritis.^(49,50) Moreover, this observation is in line with findings that FOXO1 also plays a role in angiogenesis in soft tissue wound healing.⁽⁵¹⁾

Overexpression of FOXO1 induced VEGFA. This is likely due to direct interaction of FOXO1 with the VEGFA promoter, which contains a FOXO1 consensus response element. The capacity of FOXO1 to stimulate VEGFA expression is dependent in part upon the acetylation status of FOXO1. Acetylated FOXO1 had a significantly reduced capacity to induce VEGFA promoter activity and FOXO1 did not bind to the VEGFA promoter in the presence of a deacetylase inhibitor. In contrast, deacetylated FOXO1 strongly induced VEGFA transcription activity and the binding of FOXO1 to the VEGFA promoter was enhanced by a p300 acetylase inhibitor. Both the luciferase and qPCR experiments show that altering the acetylation status of FOXO1 affects VEGF levels. These results are consistent with each other and consistent with reports that FOXO1 induced transcriptional activity is enhanced by deacetylation and reduced by acetylation.⁽⁵²⁾

It was recently reported that FOXO1 knockdown in chondrocytes by Col2α1-regulated Cre recombinase leads to increased articular cartilage thickness in mice at age 2 months, which was associated with increased cell proliferation and variable regulation of genes involved in chondrocyte differentiation.⁽⁵³⁾ We examined the impact of chondrocyte differentiation on FOXO1 expression but did not obtain consistent results (data not shown). Interestingly, it has been reported that deletion of FOXO1 alone did not cause osteoarthritis, whereas simultaneous deletion of three FOXOs (FOXO1, FOXO3, and FOXO4) resulted in premature loss of cartilage.⁽⁵³⁾ In addition, we have reported that FOXO1 regulates RANKL expression in chondrocytes in diabetic fracture healing.⁽⁵⁴⁾ These results plus those reported here indicate that FOXO1 and other FOXOs in chondrocytes play an important role in several aspects of cartilage regulation by inducing expression of several important downstream gene targets in development, fracture healing, and osteoarthritis.

In summary, our results show for the first time that chondrocytes play a critical role regulating angiogenesis during endochondral ossification in fracture healing. In experimental mice with chondrocyte-specific FOXO1 deletion there was reduced formation of blood vessels, reduced capacity of chondrocytes to induce microvascular endothelial cell tube formation in vitro, and reduced expression of VEGFA. The direct binding of FOXO1 to the VEGFA promoter and FOXO1-induced VEGFA transcriptional activity in chondrocytes provides mechanistic insight into how angiogenesis is regulated during fracture healing. Although the fracture healing model in mice closely reproduces the processes found in humans there is a possibility that they may not be identical. Nevertheless, these results provide insight into how chondrocytes can promote fracture healing, particularly in regard to angiogenesis, through a FOXO1-mediated mechanism.