FGF23 protects osteoblasts from dexamethasone-induced oxidative injury

Dexamethasone (DEX) can exert a cytotoxic effect on cultured osteoblasts. The current study explored the potential osteoblast cytoprotective effect of fibroblast growth factor 23 (FGF23). In OB-6 human osteoblastic cells and primary murine osteoblasts, FGF23 induced phosphorylation of the receptor FGFR1 and activated the downstream Akt-S6K1 signaling. FGF23-induced FGFR1-Akt-S6K phosphorylation was largely inhibited by FGFR1 shRNA, but augmented with ectopic FGFR1 expression in OB-6 cells. FGF23 attenuated DEX-induced death and apoptosis in OB-6 cells and murine osteoblasts. Its cytoprotective effects were abolished by FGFR1 shRNA, Akt inhibition or Akt1 knockout. Conversely, forced activation of Akt inhibited DEX-induced cytotoxicity in OB-6 cells. Furthermore, FGF23 activated Akt downstream nuclear-factor-E2-related factor 2 (Nrf2) signaling to alleviate DEX-induced oxidative injury. On the contrary, Nrf2 shRNA or knockout almost reversed FGF23-induced osteoblast cytoprotection against DEX. Collectively, FGF23 activates FGFR1-Akt and Nrf2 signaling cascades to protect osteoblasts from DEX-induced oxidative injury and cell death.

In the primary murine osteoblasts, FGF23 exerted a similar antioxidant activity, inhibiting DEX-induced superoxide accumulation ( Figure 4D), lipid peroxidation ( Figure 4E) and mitochondrial depolarization ( Figure  4F). FGF23 treatment alone did not alter oxidative levels in the osteoblasts ( Figure 4A-4F). These results clearly show that FGF23 inhibited DEX-induced oxidative stress in osteoblasts.

FGF23 activates Nrf2 signaling in osteoblasts
Studies have shown that forced activation of Akt in osteoblasts could induce Nrf2 cascade activation, inhibiting DEX-induced oxidative injury [10,39]. Thus, we tested the potential effect of FGF23 on Nrf2 signaling. Western blotting assay results, Figure 5A,  I) were pretreated with applied concentration of FGF23 (5 or 25 ng/mL) or vehicle control (PBS) for 2h, followed by dexamethasone (DEX, 1 μM) stimulation, cells were further cultured for the indicated time periods, cell viability and cell death were tested by CCK-8 (A and G) or medium LDH release (B and H) assays respectively; expression of the listed apoptosis-associated proteins was shown (C), with cell apoptosis tested by nuclear TUNEL staining (D and I). Stable OB-6 cells with the applied FGFR1 shRNA ("shFGFR1-Seq2") were pretreated with FGF23 (5 or 25 ng/mL) or PBS for 2h, followed by DEX (1 μM) stimulation, cells were further cultured for 48h, with cell viability (E) and apoptosis (F) tested similarly. Data were mean ± standard deviation (SD, n=5). "Veh" stands for vehicle control for DEX. * p<0.05 vs. "Veh" cells with PBS pretreatment. # p<0.05 vs. DEX-treated cells with PBS pretreatment. Each experiment was repeated three times and similar results were obtained. Bar=100 μm (D). AGING demonstrated that Nrf2 protein levels were significantly elevated in FGF23-treated OB-6 cells. The mRNA expression of Nrf2-dependent genes, including HO1, NQO1 and GCLC, was significantly increased following FGF23 treatment in OB-6 cells, with Nrf2 mRNA levels unchanged ( Figure 5B). HO1 and NQO1 protein levels were increased as well ( Figure 5A). Additionally, the NQO1 activity was enhanced in FGF23-treated in OB-6 cells ( Figure 5C). These results indicate that FGF23 activated Nrf2 signaling cascade, leading to stabilization of Nrf2 protein, expression of Nrf2-pathway genes and an increase of NQO1 activity in OB-6 cells.
In the primary murine osteoblasts FGF23 treatment induced Nrf2 protein stabilization ( Figure 5G), increased mRNA expression of Nrf2-dependent genes (HO1, NQO1 and GCLC) ( Figure 5G and 5H), as well as NQO1 activity ( Figure 5I). Taken together, these results show that FGF23 activated Nrf2 signaling in osteoblasts. cells, pre-treated with LY294002 (500 nM, 30 min pretreatment) or the vehicle control (0.1% DMSO), as well as the stable OB-6 cells with the CRISPR/Cas9-Akt1-KO construct ("ko-Akt1"), were treated with FGF23 (25 ng/mL), after 10 min expression of the listed proteins in total cell lysates were shown (A). Alternatively two hours after the FGF23 treatment, cells were treated with dexamethasone (DEX, 1 μM) or the vehicle control ("Veh"), cell viability and cell death were tested by CCK-8 (B) or medium LDH release (C) assays, respectively. The stable OB-6 cells with the constitutively-active Akt1 construct (caAkt1) or the empty vector ("Vec") were subjected to Western blotting assays to test listed proteins (D). Cells were treated with DEX (1 μM) or the vehicle control ("Veh"), after 48h cell viability (E) and cell death (F) were tested. The primary murine osteoblasts were pretreated with LY294002 (500 nM, 30 min pretreatment), followed by FGF23 (25 ng/mL) treatment for 2h, cells were further stimulated with DEX (1 μM) for 48h, cell viability (G) and death (H) were tested. Data were mean ± standard deviation (SD, n=5). # p<0.05. Each experiment was repeated three times and similar results were obtained.

DISCUSSION
The results of this study suggest that the functional FGFR1 is expressed in OB-6 osteoblastic cells and primary murine osteoblasts. FGF23 treatment in vitro induced phosphorylation of FGFR1 and its downstream Akt-S6K1 in OB-6 cells and murine osteoblasts. FGFR1 silencing in OB-6 cells, by targeted shRNAs, largely inhibited FGF23-induced Akt-S6K1 phosphorylation, but augmented with ectopic overexpression of FGFR1. treated with FGF23 (25 ng/mL) for 4h, expression of listed genes in total cell lysates was tested by Western blotting and qPCR assays (A, B, G  and H), with relative NQO1 activity tested as well (C and I). The control OB-6 cells, pre-treated with LY294002 (500 nM, 30 min pretreatment), as well as the stable OB-6 cells with the lentiviral FGFR1 shRNA ("shFGFR1") or the CRISPR/Cas9-Akt1-KO construct ("ko-Akt1"), were treated with FGF23 (25 ng/mL) for 4h, expression of HO1 and NQO1 mRNA was shown (D and E). The relative HO1 and NQO1 mRNA expression in stable OB-6 cells with the constitutively-active Akt1 construct (caAkt1) or the empty vector ("Vec") was tested (F). Data are presented as the mean ± standard deviation (n=5).* p<0.05 vs. PBS treatment. # p<0.05. Each experiment was repeated three times and similar results were obtained. AGING Importantly, FGF23-induced osteoblast cytoprotection against DEX is mediated by FGFR1-Akt signaling. With FGFR1 silencing (by shRNA), Akt inhibition (LY294002) or Akt1 KO (by CRISPR/Cas9d) FGF23 was ineffective against DEX-induced cell death and apoptosis. Conversely, forced Akt activation, by caAkt1, mimicked FGF23's action and inhibited DEXinduced cytotoxicity in osteoblasts. These results suggest that FGF23 activated FGFR1-Akt signaling to protect osteoblasts from DEX-induced cell death and apoptosis.
However the uses of traditional Nrf2 activators are limited due to their high-concentrations and possible off-target toxicities. In the present study we showed that cells with the lentiviral Nrf2 shRNA ("shNrf2) or the lentiCRISPR-GFP-Nrf2 KO construct ("ko-Nrf2"), as well as the parental control cells ("Ctrl") were treated with FGF23 (25 ng/mL) for applied time periods, expression of listed mRNAs and proteins was shown (A and B); Cells were pretreated for 2h with FGF23 (25 ng/mL), followed by dexamethasone (DEX, 1 μM) stimulation for 48h, cell viability and death were tested by CCK-8 (C) and medium LDH release (D) assays, respectively. Stable OB-6 cells with CRISPR/Cas9-Keap1-KO construct ("ko-Keap1") were treated with or without FGF23 (25 ng/mL) for 4h, control cells were tranduced with empty vector ("Vec"), expression of listed proteins was shown (E). Alternatively, cells were pretreated for 2h with FGF23 (25 ng/mL), followed by dexamethasone (DEX, 1 μM) stimulation for 48h, cell viability (F) and death (G) were tested. Data are presented as the mean ± standard deviation (n=5). # p<0.05. The experiments in this figure were repeated three times, and similar results were obtained. AGING FGF23, at only ng/mL concentrations, induced significant Nrf2 cascade activation, causing Nrf2 protein stabilization, expression of Nrf2-pathway genes and an increase of NQO1 activity in OB-6 cells and primary murine osteoblasts. Functional studies showed that FGF23 efficiently attenuated DEX-induced oxidative injury in osteoblasts, suppressing superoxide accumulation, lipid peroxidation and mitochondrial depolarization. Therefore, Nrf2 signaling activation by FGF23 exerted anti-DEX osteoblast cytoprotection in osteoblasts.
Studies have implied that Akt (and its downstream mTOR) could be an important upstream molecule of Nrf2 signaling cascade. Lee et al., found that Nrf2 activation by sulforaphane was dependent on activation of the upstream PI3K-Akt [45]. Xu et al., have shown that PI3K-Akt activation is required for pyocyanin-induced Nrf2 activation [46]. Zhang et al., demonstrated that salvianolic acid A (Sal A)-activated Akt phosphorylated Nrf2 at Ser-40 in retinal pigmentation epithelial (RPE) cells, causing Nrf2 protein stabilization and activation [47]. Here we discovered that Akt activation mediated FGF23induced Nrf2 signaling activation in osteoblasts. LY294002, the Akt inhibitor, or Akt1 knockout abolished FGF23-induced expression of HO1 and NQO1 in OB-6 cells. Contrarily, HO1 and NQO1 expression was significantly increased in OB-6 cells with caAkt1. These results indicated that FGF23induced Akt activation severed as the upstream signaling for Nrf2 cascade activation in OB-6 cells.

CONCLUSION
Collectively, these results suggest that FGF23 activates FGFR1-Akt signaling cascade to protect osteoblasts from DEX-induced oxidative injury and cell death.

Cell culture
OB-6 human osteoblastic cells [33] and primary murine osteoblasts were differentiated and cultured as described previously [33,48]. The protocols of this study were approved by IACUC and Ethics committee of Nanjing Medical University. Osteoblast cell differentiation was induced by changing to media containing 10% FBS supplemented with BMP-4 (100 ng/mL).

Western blotting
After treatment of cells, total cellular lysates (30-40 μg per lane) were separated by 10% SDS-PAGE gels, thereafter transferred onto polyvinylidene difluoride (PVDF) blots. Afterwards, the blots were blocked and subsequently incubated with the applied primary and secondary antibodies. Enhanced chemiluminescence (ECL) reagents (Pierce, Shanghai, China) were utilized to visualize the targeted bands (based on the molecular weights) using x-ray films [13][14][15]. An ImageJ software (from NIH) was utilized for data quantification.

Forced FGFR1 overexpression
The full-length FGFR1 cDNA was synthesized by Shanghai Genechem Co. It was inserted into a GV369 construct (Shanghai Genechem Co., China). The construct and the lentivirus-packing plasmids (psPAX2 and pMD2.G, Shanghai Genechem Co., China) were co-transfected to HEK-293T cells, establishing FGFR1expressing lentivirus ("lv-FGFR1"). The viruses were enriched, filtered, and added to cultured OB-6 cells (cultured in complete medium with polybrene) for 24h. Puromycin was added in the complete medium for 6 days to select stable cells. FGFR1 over-expression in the resulting stable cells was verified by Western blotting.

CCK-8 viability
Cells were initially seeded into the 96-well plates at 4, 000 cells per well. After the applied DEX treatment, a Cell Counting Kit-8 (CCK-8, Dojindo Laboratories, Kumamoto, Japan) assay kit [49] was utilized to examine the cell viability, with CCK-8 optic density (OD) values tested at 450 nm.

Lactate dehydrogenase (LDH) assay
As described previously [13][14][15], following the applied DEX treatment cell death was tested by the LDH assay, using a commercial available two-step LDH assay kit (Takara, Tokyo, Japan). The medium LDH contents were always normalized to the total LDH contents.

TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) staining
OB-6 cells or the primary murine osteoblasts were initially seeded into the 24-well tissue-culture plates (at 1.2 x 10 4 cells per well). Following the applied DEX treatment, a TUNEL In Situ Cell Death Detection Kit (Roche Diagnostics Co., Shanghai, China) was applied to quantitatively test cell apoptosis [50]. Cells were costained with TUNEL and DAPI, visualized under a confocal fluorescent microscope (Leica, Shanghai, China). For each treatment at least 600 cells in six random views (1×200 magnification) were included to calculate TUNEL ratio (% vs. DAPI).

Mitochondrial depolarization
In cells with mitochondrial depolarization the fluorescence dye JC-1 shall aggregate in the mitochondria, forming green monomers [51]. OB-6 cells or the primary murine osteoblasts were initially seeded into the 24-well tissue-culture plates. Following the applied DEX treatment, cells were stained with JC-1 (3.5 μg/mL, Sigma), washed and tested under a fluorescence spectrofluorometer (F-7000, Hitachi, Japan) at test-wavelength of 545 nm (green).

Quantitative real-time PCR (qPCR)
The total cellular RNA (extracted through the Trizol reagents) were reverse-transcribed (RT). qPCR was performed by the SYBR green kit under the ABI-7600 fast-PCR system (Applied Biosystems). The 2 ΔΔCt method was utilized to quantify expression of targeted mRNAs, using GAPDH as the internal control. All the primers for Nrf2 pathway genes were provided by Dr. Jiang at Nanjing Medical University [52][53][54].

Superoxide detection
OB-6 cells or the primary murine osteoblasts were initially seeded into the 96-well tissue-culturing plates (4 × 10 3 cells per well). Following the applied DEX stimulation, a superoxide colorimetric assay kit (BioVision, Shanghai, China) was utilized to examine the cellular superoxide contents using the attached protocols, with the superoxide's absorbance tested at the 450 nm [52].

Lipid peroxidation
As reported previously [52] OB-6 cells or the primary murine osteoblasts were seeded into six-well plates (1 × 10 5 cells per well). Following the applied DEX stimulation, the lipid peroxidation assay kit (Abcam, Shanghai, China) was utilized to examine and quantify cellular lipid peroxidation levels, tested by the thiobarbituric acid reactive (TBAR) concentration using the described protocols [52,55].

NQO1 activity
Testing the relative NQO1 activity in osteoblastic cells or murine osteoblasts with or without DEX treatment was described previously, using menadione as the substrate [56]. NQO1 activity was always normalized to that of untreated control cells.

Constitutively-active mutant Akt1
The recombinant adenovirus constitutively-active Akt1 (caAkt1, S473D) construct was provided by Dr. Zhang [36]. Ad-caAkt1 virus or the empty vector (Ad-GFP) virus was added to the cultured OB-6 cells. GFP sorting was performed to select stable cells. caAkt1 expression in the selected single stable cells was verified by Western blotting.

Akt1 knockout
The lenti-CRISPR-GFP Akt1-KO construct was provided by Dr. Zhang at Soochow University [35]. OB-6 cells were cultured in six well plates, transfected with CRISPR/Cas9 Akt1-knockout construct. Stable cells were selected via GFP FACS sorting. Akt1 knockout (KO) in the single stable cells was verified by Western blotting.

Keap1 knockout
The lentivirus with Keap1 CRISPR/Cas9 KO plasmid was provided by Dr. Liu at Jiangsu University [57], added to cultured OB-6 cells in polybrene medium. After 48h, cells were cultured in puromycin (2.0 μg/mL)-containing medium to establish the monoclonal stable cells, where Keap1 KO was verified by Western blotting.

Statistical analysis
Experiments were repeated at least three times throughout the study. Quantitative data were expressed as mean ± standard deviation (SD). Statistics were analyzed by two-way ANOVA using a Scheffe's f-test (SPSS 21.0). To test significance between two treatment groups, a two-tailed unpaired T test (Excel 2007) was utilized. p values < 0.05 were considered statistically significant.