Fluocinolone acetonide is a potent synergistic factor of TGF-β3-associated chondrogenesis of bone marrow-derived mesenchymal stem cells for articular surface regeneration

Abstract

Articular cartilage repair remains a challenging problem. Based on a high-throughput screening and functional analysis, we found that fluocinolone acetonide (FA) strongly enhances transforming growth factor beta 3 (TGF-β3)-mediated chondrogenic differentiation of human bone marrow-derived mesenchymal stem cells (hBMSCs). In an in vivo cartilage defect model in knee joints of immunocompromised mice, transplantation of FA/TGF-β3-treated hBMSCs could completely repair the articular surface. Analysis of the intracellular pathways revealed that FA enhanced TGF-β3-induced phosphorylation of Smad2 and Smad3. Additionally, we performed a pathway array and found that FA activates mTORC1/AKT pathway. Chemical inhibition of mTORC1 with rapamycin substantially suppressed FA effect, and inhibition of AKT completely repressed chondrogenesis of hBMSCs. Inhibition of glucocorticoid receptor with mifepristone also suppressed FA effect, suggesting that FA involves binding to glucocorticoid receptor. Comparative analysis with other glucocorticoids (triamcinolone acetonide (TA) and dexamethasone (DEX)) revealed the unique ability of FA to repair articular cartilage surgical defects. Analysis of intracellular pathways showed that mTORC1/AKT pathway and glucocorticoid receptor was highly activated with FA and TA, but to a less extent with DEX. Collectively, these results show a unique ability of FA to enhance TGF-β3-mediated chondrogenesis, and suggest that the FA/TGF-β3 combination may be used as major inducer of chondrogenesis in vitro. Additionally, FA/TGF-β3 could be potentially applied in a clinical setting to increase the efficiency of regenerative approaches based on chondrogenic differentiation of stem cells.

1- Introduction

Degenerative joint disorders, such as osteoarthritis or rheumatoid arthritis, affect millions of individuals worldwide and cause significant reduction in the quality of life due to compromised joint mobility and pain (1). A great number of reports have described attempts to regenerate articular cartilage by the means of cell transplantation (2), growth factors (3, 4), gene therapy (5) or chemical compounds (6, 7). However, the results achieved have not been optimal due to a certain degree of inefficiency in the methods and the limited ability of cartilage to regenerate. Recent attempts to take advantage of induced pluripotent stem cells to regenerate articular cartilage have not yet been successful (8).

Members of the transforming growth factor-beta superfamily (e.g., TGF-βs, bone morphogenetic proteins (BMPs)) and insulin-like growth factor (IGF) have been regarded as the main inducers of chondrogenesis (3). There are 3 isoforms of TGF-β (TGF-β1, TGF-β2 and TGF-β3), among which, the latter has been reported to have stronger effect in inducing chondrogenesis of human bone marrow stem/progenitor cells (hBMSCs) (9). TGF-βs activate Smad2 and Smad3, which subsequently binds to co-Smad4 and act as transcription factors or activators in the nucleus (10). BMP-2 is the major representative isoform of BMPs used in osteogenic and chondrogenic differentiation of hBMSCs, and its function is mainly through activation of Smad1/5/8. IGF family comprises IGF-I and IGF-II, which are also reported to be important factors inducing chondrogenesis of hBMSCs, together with insulin. Both insulin and IGFs activate insulin receptor and downstream pathways; however, insulin was shown to be more selective as a chondrogenic differentiating agent, whereas IGF-1 was shown to induce more proliferating effects (11). A combination of these growth factors would possibly improve the efficiency of chondrogenic differentiation of hBMSCs. Nevertheless, a previous study reported that BMP-2 and TGF-β1 can have antagonistic effects on chondrocyte proliferation and differentiation in vivo (12). Additionally, concomitant in vivo delivery of IGF-I and TGF-β1 also induced antagonistic effects compared to the effect of each factor alone (13, 14). Therefore, further investigation on the detailed mechanism of the interactions between growth factors in the context of chondrogenesis is still necessary.

Bioactive molecules have been used as important tools to promote stem cell differentiation (15), including toward the chondrogenic lineage (6). However, functional analysis of the interaction between bioactive molecules and growth factors has not been studied. In this study, we hypothesized that small molecules in combination with growth factors could potentiate chondrogenic differentiation of mesenchymal stem/progenitor cells (MSCs). We performed a high-throughput screening with prechondrogenic cells to search for compounds that can promote chondrogenesis, and further analyzed their interaction with major growth factors used in chondrogenesis of hBMSCs (TGF-β, and BMP-2). We found that the glucocorticoid (GC) fluocinolone acetonide (FA) dramatically and specifically enhanced TGF-β3-mediated chondrogenesis of hBMSCs. Further in vivo experiments confirmed the strong and unique ability of FA, compared to other glucocorticoids, in promoting the repair of surgical defects in articular cartilage.

2. Material and Methods

2.1. Recombinant proteins, antibodies and chemicals

Recombinant human TGF-β3 was purchased from R&D systems (Minneapolis, MN, USA). Recombinant human BMP-2 was purchased from PeproTech (Rocky Hill, NJ, USA). Antibodies against lubricin and glucocorticoid receptor (GR: phosphorylated-GR-S211 and total) were purchased from Abcam (Cambridge, UK). Antibodies against Smad-2, Smad-3, AKT, ribosomal protein p70S6k and their respective phosphorylated epitopes were purchased from Cell Signaling Technology Inc. (Danvers, MA, USA). Antibodies against SOX9 and collagen type II were purchased from Chemicon (Billerica, MA, USA). Anti-β-actin antibody and dexamethasone (DEX) were purchased from Sigma (St Louis, MO, USA). An FDA-approved drug library and orphan ligand drug library were purchased from Enzo Life Sciences (Tokyo, Japan). Fluocinolone acetonide (FA), triamcinolone acetonide (TA) and mifepristone/RU486 were purchased from Tokyo Chemical Industry Inc. (Tokyo, Japan). Rapamycin and AKT1/2 inhibitor vii were purchased from Calbiochem (San Diego, CA).

2.2. Cells and culture conditions

ATDC5 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) and Ham’s F-12 (DMEM/F-12; Invitrogen, Carlsbad, CA, USA) medium containing 5% fetal bovine serum (FBS: Invitrogen), 1% penicillin and streptomycin (Sigma) and 1% L-glutamine (Invitrogen). hBMSCs were purchased from Lonza (Walkersville, MD, USA) and cultured in alpha-MEM (Invitrogen) containing 15% FBS (Invitrogen), 100 mM L-ascorbic acid 2-phosphate (Wako Pure Chemical Industries, Osaka, Japan) and 1% antibiotics.

For chondrogenic differentiation, ATDC5 cells or hBMSCs (at passage 3 to 6) were suspended at a concentration of 2 × 10⁷ cells/mL and cultured in 10 μL micromasses, according to a previously reported protocol (16). Basic (control) chondrogenic medium consisted of DMEM (low glucose) containing 6 μg/mL of insulin, transferrin and selenous acid (ITS solution, BD, Bedford, MA, USA), L-ascorbic acid (50 μg/mL) and antibiotics. For induction experiments, the medium was supplemented with TGF-β3 (5 ng/mL) and/or compounds/drugs or vehicle (dimethyl sulfoxide, DMSO, Sigma). Cultures were maintained for 21 days, with the medium changed every 3 to 4 days, and further submitted to analysis of mRNA levels or histology. Analysis of SOX-9 protein levels was performed in micromasses cultured for only 1 day.

2.3. In vivo experiments

Severe combined immunodeficient (SCID) nude mice (Balb/c, CLEA, Japan Inc., Tokyo, Japan) at 8 weeks of age were utilized for the experiments according to the Guidelines for Animal Research of Okayama University Dental School as well as the principles of the Declaration of Helsinki. Before surgery, general anesthesia was induced by inhalation of isoflurane (Isoflu: Dainippon Sumitomo Pharma Co., Osaka, Japan). Knee joints were disinfected with 70% ethanol and exposed by a vertical incision at the medial portion of the knee. Patella was dislocated laterally after incision, and full-thickness cartilage defects were manually performed in the center of the joint by a single investigator (E.S.H.) with a dental steel round bur #1/2 (Drendel & Zweiling, Berlin, Germany) of 500 μm in diameter. hBMSCs pellets cultured previously in chondrogenesis-inducing medium with or without factors for 5 days, were cut in half with a surgical blade and transplanted into the surgical defect. Muscles and patella were replaced on their original position and sutured with a degradable suture thread (Opepolix, Alfressa Pharma, Osaka, Japan). The outer skin was then sutured with nylon thread (Softretch, GC, Tokyo, Japan). Mice in the sham-operated group had only the patella dislocated. The mice were euthanized one month after the operation for studies of hindlegs and knee joints. The research protocol for animal experiments was approved by the Committee for animal care and use of Okayama University (OKU- 2013513). The study also conformed with the ARRIVE guidelines for preclinical procedures.

2.4. Histological analysis and qualitative scoring

hBMSC pellets cultured for 21 days were fixed in 4% paraformaldehyde (PFA), dehydrated and embedded in paraffin. Serial sections of 5 μm were made, hydrated, and subsequently stained with hematoxylin and eosin (H&E), or toluidine blue and safranin O for staining of glycosaminoglycan, as described previously (17).

Knee joint samples were fixed in 4% PFA, decalcified in Morse’s solution or 10% ethylene diamine-tetra acetic acid (EDTA) solution, dehydrated and embedded in paraffin. Specimens were sectioned sagittally in approximately 100 serial sections of 5 μm and mounted on silane-coated slides. Representative sections of each specimen were stained with H&E, or toluidine blue for further histological scoring. In a blinded fashion, three examiners used a modified ICRS score (Suppl. Table S2) to evaluate the histological condition of the newly-formed articular cartilage surface. Immunohistochemical (IHC) analysis for collagen II and lubricin was performed according to previous methods. (18, 19)

2.5. Transfection and dual-luciferase reporter assay

For library screening, electroporation was used to transfect ATDC5 cells with luciferase reporter plasmids encoding a collagen type II alpha 1 (Col2a1) promoter fragment or Renilla, as reported previously (20). Measurements of Col2a1-driven luciferase activity were performed after an incubation period of 72 h with 640 FDA-approved drugs or 84 orphan ligands. For analysis of Smad activity, hBMSCs were seeded in 96-well plates, and after 24 h transfected with CAGA reporter plasmids and Renilla using lipofectamine 2000 (Invitrogen), according to manufacturer’s instructions. Eight-hours after transfection, the cells were stimulated with FA and/or TGF-β3, and luciferase activity was measured after 30 min.

2.6. Real time reverse-transcription polymerase chain reaction (RT-PCR) analysis

For the second screening step, total cellular RNA was extracted using the RNeasy 96 kit after ATDC5 cells were stimulated with drugs for 24 h. For other experiments, RNeasy mini kit (QIAGEN, Hilden, Germany) or Purelink (Invitrogen) was used, according to the manufacturer’s instructions. To remove potential residual DNA the samples were treated with DNase I (QIAGEN RNase-Free DNase Set; or DNASE 1, Invitrogen). Real time RT-PCR was used for mRNA quantitation as described (20). The levels of mRNA of interest were normalized to that of the reference gene ribosomal protein S29. Primer sequences are shown in Supplementary Table S3. All experiments were repeated at least three times, independently.

2.7. Pathway array and western blot analysis

Total cellular protein was extracted with a mammalian protein extraction reagent (Thermo Scientific, Waltham, MA, USA), according to the manufacturer’s instructions; protease inhibitors (Roche, Basel, Switzerland) and phosphatase inhibitor cocktail (Thermo Scientific) were added. For analysis of phosphorylated proteins, protein samples were collected 10 min after stimulation with drugs or inhibitors. Protein concentrations were determined using BCA protein assay kit (Thermo Scientific), according to manufacturer’s instructions. Ten or twenty micrograms of protein were resolved on precast gels (NuPAGE Novex 4–12% Bis-Tris Gel; Invitrogen), transferred onto a PVDF membrane, and immunoblotted with antibodies, as reported previously (20). Proteins were visualized using Luminata Forte Western HRP substrate (EMD Millipore, Billerica, MA, USA) or SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific). All experiments were repeated at least three times, independently. A pathway array screen was performed once using a commercially available product (PathScan RTK Signaling Antibody Array Kit, Cell Signaling).

2.8. Immunocytochemistry

Immunofluorescence staining of cells after 10 min, 30 min and 60 min of treatment with glucocorticoids was performed using first antibody against glucocorticoid receptor (GR) phosphorylated at serine 211 (GR-S211) (Abcam) or IgG and second antibody AlexaFluor 488 conjugated IgG (Invitrogen), according to a previous method (21). Nuclei were stained with 4′, 6-Diamidino-2-Phenylindole (DAPI, Sigma). For quantitative analysis an automated fluorescence imaging system (Cellomics ArrayScan VTI high content screening reader, Thermo Scientific) was used.

2.9. Statistical analysis

Analysis of the differences between groups were performed with unpaired Student’s t-test, or one-way ANOVA followed by a Tukey post-hoc correction test when appropriate. GraphPad Prism 5 software was used for the analyses.

3. Results

3.1. FA in combination with TGF-β3 potentiates chondrogenesis and promotes cartilage repair

In an attempt to discover drugs that can promote chondrogenesis, we screened a Food and Drug Administration (FDA)-approved drug library containing 640 compounds (Suppl. Table S1) and an orphan ligand library containing 84 compounds (20) using a pre-chondrogenic cell line (ATDC5 cells) transfected with a Col2a1 promoter-driven luciferase reporter in monolayer culture (Fig. 1a). This first screen identified 86 compounds that induced at least a two-fold increase in Col2a1 promoter activity. In a second screening, we analyzed the mRNA levels of aggrecan (Acan), which is another major marker of chondrocytes, using a high-throughput RNA quantification system (Fig. 1b). Only 8 compounds up-regulated Acan expression levels, and these were submitted to a third screening assay for confirmation of the mRNA levels of Acan, as well as those of the master regulator of chondrogenesis, Sox9 (Fig. 1c). This third assay identified FA as a compound able to significantly increase Acan and Sox9 mRNA levels (Fig. 1c). The increase in Sox9 was also confirmed at a protein level in ATDC5 cells (Fig. 1d).

Fig. 1. Screening process using ATDC5 cells.

(a) Relative luciferase reporter activity of firefly and renilla luciferase after 72 h of stimulation with 640 FDA-approved drugs and 84 orphan ligands, compared to negative control. Eighty-six drugs/compounds induced more than two-fold increase in Col2a1 promoter activity. Experiments were performed in duplicate. (b) Second screening showing Acan mRNA levels after 1 day of stimulation with 86 compounds selected in the first screen. Only 8 compounds induced more than 1.5 fold increase in Acan levels. Experiments were performed in duplicate. RNA could not be isolated from samples number 26 and 27 because of the toxicity induced by the drugs daunorubicin-HCl and doxorubicin-HCl, respectively. For (a–b), asterisks identify FA. (c) Results of the third screening showing that FA was the only drug inducing an increase in mRNA levels for Sox9 and Acan. Experiment was performed once. (d) Western blot analysis confirming the FA-induced increase in Sox9 levels. For the screening process, ATDC5 cells were cultured in monolayer. Concentrations of all drugs in all screening steps were 2 μg/mL.

Next, in order to investigate the interaction of FA with other major factors involved in chondrogenic differentiation, we cultured hBMSCs in micromass cultures for 21 days in basic chondrogenic medium (insulin, transferrin, selenous acid and ascorbic acid) with or without TGF-β3 (5 ng/mL) and/or FA (2 μg/mL, 4.4 μM). Basic chondrogenic medium was sufficient to induce chondrogenic differentiation of hBMSCs, mainly due to insulin action (11). However, TGF-β3 supplementation resulted in a decreased deposition of cartilaginous matrix as shown by safranin O staining of hBMSC pellets (Fig 2a). A possible explanation for these findings could be associated with the low amount of TGF-β3 used in the experiments, or antagonistic effects between insulin and TGF-β3, as reported in previously (13, 14). Of note, contrary to the results obtained in monolayer culture, FA alone suppressed chondrogenesis in these three-dimensional cultures (Fig. 2a; Suppl. Fig. S1a, b). However, in combination with TGF-β3, FA markedly potentiated chondrogenesis, as indicated by strong staining of glycosaminoglycans with safranin O, and immunohistochemical staining for collagen type II in sections of hBMSC pellets (Fig. 2a). Consistently, the FA/TGF-β3 combination induced up-regulation of SOX9 levels (Fig. 2b), indicating enhanced commitment of hBMSCs towards the chondrogenic lineage. A quantitative analysis of mRNA levels in the pellets also showed a strong up-regulation of mature chondrocyte markers, ACAN and COL2A1, upon FA/TGF-β3 stimulation (Fig. 2c). Analysis of the transcript levels of the hypertrophic chondrocyte marker, COL10A1, showed only a moderate increase, which was not proportional to the levels of COL2A1, but could reflect a normal process of chondrocyte differentiation of the cells (Fig. 2c).

Fig. 2. Potent chondrogenic differentiation of hBMSCs by FA/TGF-β3.

(a) Histological analysis of hBMSC micromass cultures. FA in combination with TGF-β3 strongly enhances glycosaminoglycan deposition and collagen II synthesis as shown by intense staining with safranin O and immunohistochemical analysis, respectively. Control refers to the group cultured in basal chondrogenic medium. (b) FA/TGF-β3 dramatically induces SOX9 levels. Blot images are representative of three different experiments. (c) Analysis of mRNA levels of ACAN, COL2A1 and COL10A1 of 21-day hBMSC pellets. * p<0.05, *** p<0.001, ANOVA/Tukey test (N = 3). Values are mean ± s.e.m. For (a–b), vehicle (DMSO) was added to control and TGF-β3 groups. Concentrations of FA and TGF-β3 were 2 μg/mL and 5 ng/mL, respectively. (d) Histological analysis of 21-day hBMSC micromass cultures showing that FA inhibited BMP-2-associated chondrogenic differentiation of hBMSCs. Note the great deposition of glycosaminoglycan as detected by safranin O staining only in the group stimulated with FA/TGF-β3. (e–f) FA/BMP-2 stimulation suppressed SOX9 protein levels as well as ACAN, COL2A1 and COL10A1 mRNA levels. For (d–f), TGF-β3, BMP-2 and FA concentrations were 5 ng/mL, 100 ng/mL and 100 nM, respectively. Squares indicate the area shown in high magnification images. *** p<0.001, ANOVA/Tukey test (N = 3). Values are mean ± s.e.m.

Since bone morphogenetic protein 2 (BMP-2) has also been reported to play crucial roles in cartilage development, we also examined whether FA could positively interact with BMP-2. Interestingly, FA suppressed BMP-2-mediated chondrogenesis of hBMSCs as demonstrated by analysis of SOX9 protein levels, mRNA levels of ACAN and COL2A1 and histological staining of 21-day hBMSC micromasses (Fig. 2d–f). These results demonstrate that FA potentiates chondrogenesis of hBMSCs via a specific synergistic interaction with TGF-β3.

We next investigated whether the combination of FA/TGF-β3 could be used to effectively repair articular cartilage defects in vivo by transplanting hBMSC micromasses cultured for 5 days in the presence of FA, TGF-β3 or a combination of both, into full-thickness articular cartilage defects in the knee joints of SCID/nude mice. A model of this cartilage defect had been validated before the experiments were performed (Fig. 3a, Suppl. Fig. S2). There was no repair of the articular cartilage in control or TGF-β3 groups, although a chondrocyte cluster could be observed in the area underlying the articular surface (Fig. 3b). As expected from the results observed in vitro, FA-treated hBMSCs induced no new cartilage formation. Interestingly, however, the combination of FA/TGF-β3 markedly promoted repair of the defect resulting in a smooth surface of the newly-formed articular cartilage, close to that of the sham-operated group. Immunohistochemical analysis showed expression of collagen type II throughout most of the repaired cartilage, whereas in control, TGF-β3 and FA group, it was localized in areas of aggregated chondrocytes under the region corresponding to the articular cartilage. Lubricin, a protein that lubricates the cartilage, was detected in chondrocytes and extracellular matrix at the articular surface in FA/TGF-β3 and sham group, whereas in control group, lubricin was found in the chondrocyte cluster under the superficial zone. Histopathology-based scoring of the repair cartilage showed a significant difference between the FA/TGF-β3 group and all other groups, except for the sham-operated group (Fig. 3c).

Fig. 3. FA/TGF-β3 promotes the articular surface regeneration.

(a) Full-thickness articular cartilage defect model in the knee of mice. (b) Histological analysis of mouse knee joints showing substantial articular cartilage repair in the group transplanted with FA/TGF-β3-treated hBMSCs, as demonstrated by a smooth articular surface stained with toluidine blue. Note the presence of collagen II in the surface of the newly-formed articular cartilage. Control refers to the group that was transplanted hBMSCs cultured in basic chondrogenic medium. Arrowheads show the borders of the surgical defect. Asterisks show the region corresponding to the superficial layer of the articular cartilage which could not be repaired in any group, except for FA/TGF-β3 group. Red arrows indicate chondrocyte clusters below the region corresponding to the articular cartilage surface. Black arrows indicate cells or extracellular matrix regions where lubricin was detected. (c) Histological assessment of knees showing a significant difference among FA/TGF-β3 and all other experimental groups, except for the sham-operated group. N = 5/group, in two independent experiments. ** p<0.01, *** p<0.001, ANOVA/Tukey test. Values are mean ± s.e.m.

3.2. Intracellular mechanisms involved in FA/TGF-β3-induced chondrogenesis of hBMSCs

In an attempt to identify the intracellular mechanism behind the FA effect, we first investigated whether FA stimulates TGF-β3 signaling. Western blot analysis indicated that FA enhanced TGF-β3-induced phosphorylation of both Smad2 and Smad3 (Fig. 4a), and this was further confirmed by the finding that FA stimulated a CAGA box-driven luciferase reporter (Fig. 4b). Since Smad3 activation has been reported to enhance SOX-9 and COL2a1 transcript levels (10), these results partially explain the effect of FA in potentiating TGF-β3-mediated chondrogenesis of hBMSCs.

Fig. 4. FA enhances TGF-β3-induced phosphorylation of Smads and activates mTORC1/AKT signaling pathway.

(a) FA enhanced TGF-β3-induced phosphorylation of Smad2 and Smad3 at 10 min after stimulation. (b) CAGA-driven luciferase reporter assay indicating enhanced Smad activity at 30 min after FA/TGF-β3 stimulation. Assays were repeated at least three times, independently. * p<0.05, ** p<0.01, ANOVA/Tukey test. RLA = relative luciferase activity. (c–d) Western blot analyses showing that FA alone or in combination with TGF-β3 induces phosphorylation of p70S6K, a direct downstream target of mTORC1, and AKT. Rapamycin and AKT1/2 inhibitor completely inhibited FA-induced activation of mTORC1 and AKT, respectively. (e) SOX9 levels were partially blocked with rapamycin, and completely suppressed with AKT inhibitor after FA/TGF-β3 stimulation. (f, g) Rapamycin inhibited FA/TGF-β3-mediated chondrogenesis as shown by suppression of glycosaminoglycan deposition (f) and of mRNA levels of ACAN, COL2A1 and COL10A1 (g). AKT1/2 inhibitor completely suppressed chondrogenesis, as staining with safranin O (f) and mRNA levels of COL2A1 could not be detected (g). Images are representative of three different experiments. Rapa = rapamycin. AKT_inhib = AKT1/2 inhibitor. Phosphorylated = P. *** p<0.001, ANOVA/Tukey test (N = 3). Values are mean ± s.e.m. Scale bars, 200 μm.

However, since the activation of Smads seemed not to be proportional to the increase in chondrocyte markers, we hypothesized that alternative pathways could be involved in FA effect. Therefore, we used a pathway array to measure serine/tyrosine phosphorylation of components in receptor tyrosine kinase pathways which are known to control numerous cellular functions such as division, growth, metabolism, differentiation and survival. The results showed a stronger phosphorylation of ribosomal protein p70S6 kinase — a direct downstream of the mammalian target of rapamycin complex 1 (mTORC1) — and phosphorylation of AKT/protein kinase B (PKB) at serine 473 after stimulation with FA/TGF-β3 (Suppl. Fig. S3). Subsequent western blot assays with hBMSCs confirmed that FA induced activation of both p70S6k/mTORC1 and AKT, and this was completely blocked by the mTORC1 inhibitor, rapamycin, and an AKT1/2 inhibitor (Fig. 4c, d).

In order to confirm the involvement of these pathways in FA/TGF-β3-induced chondrogenesis of hBMSCs, we performed a functional analysis with chemical inhibition of mTORC1 and AKT during chondrogenesis of hBMSCs (Fig. 4e–g). Blockade of mTORC1 with rapamycin partially inhibited SOX9 protein levels (Fig. 4e). Consistently, rapamycin treatment decreased the staining with safranin O and toluidine blue as well as mRNA levels of ACAN, COL2A1 and COL10A1 of 21-day hBMSC pellets (Fig. 4f, g). Notably, inhibition of AKT completely repressed SOX-9 levels (Fig. 4e) as well as staining for glycosaminoglycans of the hBMSC pellets. Additionally, expression levels of COL2A1 could not be detected in the pellets, indicating that AKT signaling play a crucial role in chondrogenic differentiation of hBMSCs.

Since GCs are known to bind to cytoplasmic GR, we also performed inhibition assays with the GR inhibitor, mifepristone (RU486). As shown in Suppl. Fig. S4a–c, blockade of GR suppressed the FA/TGF-β3-induced chondrogenesis in hBMSCs, as demonstrated by reduced levels of SOX9, ACAN, COL2A1, as well as reduced deposition of cartilaginous matrix as shown by safranin O staining of hBMSC pellets. RU486 also inhibited the FA effect on TGF-β3-induced phosphorylation of Smads, and on phosphorylation of mTORC1/AKT (Suppl. Fig. S4d, e). These results suggest that the FA activity possibly occurs subsequently to its binding to GR.

3.3. Comparison of the effect of other glucocorticoids on TGF-β3-mediated chondrogenesis of hBMSCs

Next, we compared FA with other glucocorticoids with similar molecular structure, TA and DEX (Fig. 5a). Commitment of hBMSCs to chondrogenic lineage, as shown by SOX9 levels, was highly induced following FA/TGF-β3 stimulation, but only at a moderate level after TA/TGF-β3 or DEX/TGF-β3 treatment (Fig. 5b). In accordance, cartilaginous matrix deposition, as demonstrated by safranin O staining, collagen type II immunostaining, as well as by an increase in mRNA levels of ACAN and COL2A1, was evidently higher with FA/TGF-β3 compared to TA/TGF-β3 or DEX/TGF-β3, either at 100 nM (Fig. 5c, d) or 1 μM concentrations (Suppl. Fig. S5).

Fig. 5. Comparison of different glucocorticoids on TGF-β3-mediated chondrogenesis of hBMSCs.

(a) Molecular structure of FA, TA and DEX. (b) SOX9 protein levels detected by western blot upon treatment with FA, TA and DEX. (c) Histological analysis of hBMSC micromass cultures showing that FA/TGF-β3 induced strong deposition of glycosaminoglycan (safranin O staining) and collagen II synthesis compared to TA/TGF-β3 and DEX/TGF-β3. Images are representative of three different experiments. Squares indicate the area shown in high magnification images. (d) Increases in ACAN, COL2A1 mRNA levels are evidently higher following FA/TGF-β3 stimulation, compared to TA/TGF-β3 or DEX/TGF-β3 stimulation. Transcript levels of COL2A1 are shown in logarithmic scale. * p<0.05, ** p<0.01, *** p<0.001, ANOVA/Tukey test. Results are mean ± s.e.m. of three independent experiments. GC concentration was 100 nM.

Analysis of intracellular pathways showed a tendency for an enhanced phosphorylation of Smads after stimulation with TGF-β3 and the GCs (Fig. 6a, b). Additionally, activation of mTORC1/AKT was stronger after GC/TGF-β3 stimulation, compared to stimulation with TGF-β3 alone (Fig. 6c, d). We also quantified the phosphorylated GR-S211 in the nucleus, which has been correlated with GR transcriptional activity, and detected no difference in the phosphorylated GR at serine 211 (P-GR-S211) after FA or TA stimulation; whereas the efficiency of DEX to activate GR-S211 was markedly low (Fig. 6e–g; Suppl. Fig. S6).

Fig. 6. Comparison of intracellular pathways upon stimulation with different glucocorticoids.

(a) FA, TA and DEX moderately enhanced TGF-β3-induced phosphorylation of Smad2 and Smad3. (b) Relative intensity of P-Smads/total Smads blots are plotted graphically (N = 3). (c) GCs also enhanced mTORC1/AKT signaling, compared to TGF-β3 alone. (d) Relative levels of P-AKT/total AKT and P-p70S6/total p70S6 are plotted graphically (N = 3). (e) Western blot analysis of GR phosphorylation 10 min after GC stimulation. Results showed no difference in the activation of GR-S211 by FA or TA, whereas activation of GR-S211 by DEX was moderate. (f) Relative levels of P-GR/total GR are plotted graphically (N = 3). (g) A time-dependent quantitative analysis of immunocytochemical assay showing the number of cells positive for P-GR-S211 after GC stimulation. Results confirm stronger activation of GR-S211 by FA and TA. Results are mean ± s.e.m. of three independent experiments. Statistical analyses were performed with one way ANOVA in each time point independently. For (b, d, f, g), * p<0.05, ** p<0.01, *** p<0.001, ANOVA/Tukey test. GC concentration was 100 nM. P = Phosphorylated. hBMSCs in monolayer culture were stimulated with factors under basic culture medium.

Finally, the in vivo cartilage repair model confirmed that FA/TGF-β3 is uniquely able to promote cartilage repair, as demonstrated by toluidine blue staining, collagen II immunostaining and qualitative scoring of histological sections (Fig. 7). Of note, although hBMSCs pellets treated with TA/TGF-β3 presented a significant increase in chondrocyte markers in vitro, the cells could not effectively repair the articular cartilage in vivo. On the contrary, we could observe a degradation of intact articular cartilage in the group that received TA/TGF-β3-treated hBMSCs, as demonstrated by substantial decrease in toluidine blue staining in the superficial zone of the articular cartilage (Fig. 7a). This apparent harmful effect was not noticeable with FA/TGF-β3 treatment, despite the difference in a single fluorine atom between FA and TA.

Fig 7. Among glucocorticoids, FA/TGF-β3 is uniquely able to promote articular surface regeneration.

(a) Histological analysis of knee joints of mice. Substantial cartilage repair was observed only in the group transplanted with hBMSCs treated with FA/TGF-β3. Arrowheads show the borders of the surgical defect. Asterisk shows the regenerated superficial layer of the articular cartilage only in FA/TGF-β3 group. Arrows indicates articular surface damage in the groups that received TA/TGF-β3-treated hBMSCs. (b) Scoring of the histological sections revealed a significant difference only between FA and each of the other groups. N = 6/group, in two independent experiments. *** p<0.001, ANOVA/Tukey test. Values are mean ± s.e.m.

4. Discussion

The goal of this study was to screen for bioactive compounds that can itself enhance chondrogenesis and interact positively with major growth factors regulating chondrogenesis. Among a total of 724 compounds, FA was found to present unique effects in enhancing TGF-β3-mediated chondrogenesis. GCs are among various endocrine molecules that regulate cell growth and differentiation, either by acting as a transcription factor or as an activator of intracellular signaling molecules. The importance of glucocorticoids, particularly that of DEX, in chondrogenic differentiation of hBMSCs is already known (22). However, the differential effects on chondrogenesis among GCs have never been investigated. During the screening process, we compared 10 natural and synthetic GCs (Suppl. Table S1), and despite the great similarity in the molecular structure among GCs, particularly between FA and TA, the effects of GCs in inducing chondrogenic differentiation of hBMSCs were evidently different, especially in the results of the in vivo model of cartilage repair. We also found that GCs can activate intracellular pathways in a different manner; in particular, activation of GR was evidently different, which suggests that a possible reason behind the different activities of GCs could be associated with their nuclear function, probably as transcription factors or co-factors. Of note, however, we only observed a significant difference in the effects of FA, TA and DEX using micromass cultures of hBMSCs, whereas experiments with ATDC5 cells cultured under the same protocol and conditions showed no differences among the GCs (data not shown). These data provide strong support for the notion that utilization of primary cells in a tissue-like culture system is required for in vitro observations to mimic and to be relevant to in vivo mechanisms.

Additionally, since GCs are also utilized in the differentiation of stem cells into several other cell lineages (e.g., hepatocytes, myocytes), functional analysis with different GCs may reveal novel insights into the complex mechanism of how GCs affect cell differentiation in a tissue-specific manner.

Regarding the clinical application of GCs, FA is currently mainly used for dermal (23), dental (24) and ophthalmological (25) prescriptions; whereas its close counterparts, TA and DEX, have been already used for intra-articular injections in the management of joint diseases (26). Although application of GCs in rheumatoid arthritis and other joint diseases is widely adopted (27, 28), long-term application of GCs have been recognized to be harmful to cartilage (29). A possible reason for these findings could be associated with the differential effects of GCs on cartilage. Therefore, analysis of the effect of several GCs and selection of a more appropriate GC may possibly decrease the GC-induced cartilage damage.

In this study, we showed that FA activated mTORC1/AKT pathway in hBMSCs, and firstly report the effect of mTORC1 and AKT1/2 inhibition on the chondrogenic differentiation of hBMSCs. AKT pathway is one of the major pathways involved in cell survival and differentiation, and mTORC1 has been associated with protein synthesis. Therefore, FA-induced activation of mTORC1/AKT could in part explain the FA/TGF-β3 effect in enhancing chondrogenic differentiation of hBMSCs. Our data also showed that mTORC1 inhibition with rapamycin blocked the effect of FA/TGF-β3, and significantly decreased matrix synthesis and deposition in hBMSCs. On the other hand, inhibition of AKT1/2 completely repressed chondrogenic differentiation of hBMSCs, suggesting that AKT plays a crucial role possibly as a major pathway in this process. In agreement, previous investigations had also reported the importance of mTORC1/AKT pathway in chondrocytes or chondroprogenitor cells (30, 31). Rapamycin and AKT inhibitors were reported to suppress terminal differentiation of mature chondrocytes and inhibit bone growth (32–34). In the clinical setting, rapamycin and AKT inhibitors have also been utilized as a chemotherapeutic agent in the management of tumors in adults and children due to its suppressive effects on cell growth. Therefore, taking in consideration the inhibitory effect of these drugs on cartilage and bone growth, application of these drugs in the management of cancer in developing individuals should be careful and moderated.

Taken together, our data provide novel insights into interactions between growth factors and biologically active compounds that result in enhanced chondrogenic differentiation of MSCs, and suggest that FA/TGF-β3 could potentially be applied in a clinical setting to improve the outcomes of applications involving hBMSCs for cartilage regeneration.