Transcriptional activation of follistatin by Nrf2 protects pulmonary epithelial cells against silica nanoparticle-induced oxidative stress

Silica nanoparticles (SiO2 NPs) cause oxidative stress in respiratory system. Meanwhile, human cells launch adaptive responses to overcome SiO2 NP toxicity. However, besides a few examples, the regulation of SiO2 NP-responsive proteins and their functions in SiO2 NP response remain largely unknown. In this study, we demonstrated that SiO2 NP induced the expression of follistatin (FST), a stress responsive gene, in mouse lung tissue as well as in human lung epithelial cells (A549). The levels of Ac-H3(K9/18) and H3K4me2, two active gene markers, at FST promoter region were significantly increased during SiO2 NP treatment. The induction of FST transcription was mediated by the nuclear factor erythroid 2-related factor 2 (Nrf2), as evidenced by the decreased FST expression in Nrf2-deficient cells and the direct binding of Nrf2 to FST promoter region. Down-regulation of FST promoted SiO2 NP-induced apoptosis both in cultured cells and in mouse lung tissue. Furthermore, knockdown of FST increased while overexpression of FST decreased the expression level of NADPH oxidase 1 (NOX1) and NOX5 as well as the production of cellular reactive oxygen species (ROS). Taken together, these findings demonstrated a protective role of FST in SiO2 NP-induced oxidative stress and shed light on the interaction between SiO2 NPs and biological systems.

Scientific RepoRts | 6:21133 | DOI: 10.1038/srep21133 These NP-responsive proteins exert antioxidant and cytoprotective effects in lungs. If the protection effect fails, escalation of oxidative stress results in mitochondrial perturbation and cell apoptosis 1,7,8 . However, besides a few Nrf2 target genes, the regulation of other SiO 2 NP-responsive genes and their functions in SiO 2 NP response remain largely unknown.
Follistatin (FST) is widely expressed in higher animals and participates in a variety of processes such as cell growth, development, differentiation, and secretion 9 . FST was firstly identified as a secretory protein that binds and inactivates transforming growth factor (TGF)-β family members including activin, bone morphogenetic proteins (BMPs), and myostatin. The TGF-β -like molecule-neutralizing effect of FST contributes to most of its functions [10][11][12] . Recent studies showed that FST also plays a protective role under cellular stresses including oxidative stress [13][14][15] . For example, activin induces endothelial cell oxidative stress and endothelial cell dysfunction, and these effects are mitigated by follistatin 15 . Follistatin has also been reported to suppress BMP4-induced ROS production 16 . Based on the inhibitory role of FST in cellular ROS production, we proposed that FST participates in cellular response to SiO 2 NP-induced oxidative stress.
In this study, we evaluated the function of FST in SiO 2 NP-induced lung toxicity using both in vitro and in vivo models. We found that FST was transcriptionally up-regulated by transcription factor Nrf2 during SiO 2 NP treatment. Our data further demonstrated that FST protected cells against apoptosis through reversing oxidative stress. These data provided evidence to support that FST is an oxidative stress responsive protein contributing to lung anti-oxidant response to SiO 2 NPs.

Results
Characterization of SiO 2 NPs. We first evaluated the size of silica nanoparticles using TEM. Data showed that the silica nanoparticles had a spherical shape with the average diameter of 37.3 nm (Fig. 1A). Size measurement by Image J software showed that SiO 2 NPs were approximately normally distributed (Fig. 1B). The hydrodynamic size of silica nanoparticles in culture media RPMI-1640 was 41.36 ± 2.55 nm and the silica nanoparticles exhibited good monodispersity. SiO 2 NPs are quite stable in RPMI 1640 media since the size did not increase after 24 h of incubation (41.32 ± 3.19 nm, Table 1). Further, we measured zeta potentials of SiO 2 NPs, which provide quantitative information on the stability of the particles. Data showed that the absolute value of zeta potentials of SiO 2 NPs in our study was − 33.28 ± 0.65 and − 35.14 ± 0.71 mV respectively, indicating that the particles are more likely to remain dispersed. SiO 2 NP enhances FST expression in mouse lung tissue. FST is up-regulated during glucose deprivation and plays a protective role 13,14 . We hypothesized that FST is also a SiO 2 NP-responsive protein. Since lung is the major exposure tissue for SiO 2 NPs, we detected whether SiO 2 NP treatment could induce FST expression change in mouse lung tissue. To that end, we intratracheally instilled mice with SiO 2 NPs. The mRNA level of FST increased (5.7 fold) at 1 day after SiO 2 NP instillation and decreased to the basal level 4 days after SiO 2 NP treatment ( Fig. 2A, Fig. S1). Immunoblotting showed that the protein level of FST significantly increased at 1  day after SiO 2 NP treatment, while the increase of FST protein at 7 days was not as dramatic as 1 day (Fig. 2B). Furthermore, immunohistochemistry staining showed increased expression of FST in SiO 2 NP-treated lung tissue (Fig. 2C). These data indicated that FST is an early responsive gene to SiO 2 NP treatment.

SiO 2 NP enhances FST expression in A549 cells. Epithelial cells are the major components of lung
tissue and essential for lung function. We asked whether the expression of FST in lung epithelial cells increased during SiO 2 NP treatment. Thus, we treated human epithelial cells A549 with SiO 2 NPs and detected FST mRNA level. Data showed that SiO 2 NP up-regulated FST mRNA level in a dose-dependent manner. 50 μ g/ml of SiO 2 NP treatment induced 8-fold increase of FST mRNA (Fig. 2D). The mRNA level of FST started to increase 1 hour after SiO 2 NP treatment, while the protein level started to increase at 3 hours of SiO 2 NP treatment. The expression of FST reached the peak at 24 hours (Fig. 2E,F). These data indicated that SiO 2 NP promotes FST expression in lung epithelial cells. We also detected the effect of SiO2 NP in normal branchial epithelial cell line BEAS-2B. Data showed that SiO2 NP also promoted FST expression in BEAS-2B cells (Fig. S2).
The transcription of FST gene is activated during SiO 2 NP treatment. The higher level of FST mRNA may result from either transcriptional activation or post-transcriptional regulation 14 . We first examined the stability of FST mRNA. Actinomycin D was used to block de novo RNA synthesis, and then the persistence of the existing FST mRNA was measured by quantitative RT-qPCR at 1, 2, 4 and 8 h. Our results revealed that the stability of the FST mRNA did not change during SiO 2 NP treatment (Fig. 3A). Actually, when de novo RNA synthesis was inhibited, FST mRNA level could not increase during SiO 2 NP treatment (Fig. S3), suggesting that SiO 2 NP regulates FST mRNA expression at transcriptional level. Therefore, we detected changes in FST gene promoter activity using a luciferase reporter. Data showed that the transcriptional activity of FST promoter region was significantly induced under SiO 2 NP treatment (Fig. 3B). Since Ac-H3(K9/18) and H3K4me2 are recognized as active gene markers 17 , we further detected the Ac-H3(K9/18) and H3K4me2 levels at FST promoter region using chromatin immunoprecipitation. Data showed that SiO 2 NP significantly increased both Ac-H3(K9/18)  By analyzing the FST gene proximal promoter, we identified a putative antioxidant response element (ARE), which is a consensus Nrf2-binding sequence (Fig. 4D). To determine the direct regulation of Nrf2 on FST expression, the luciferase reporters containing FST promoters with either wild-type or ARE-deleted sequences were used. Data revealed that luciferase activity of the wild-type FST promoter reporter was significantly reduced (to 40% of control levels) following knockdown of Nrf2, indicating that FST promoter activity is Nrf2-dependent. However, deletion of ARE region completely abolished Nrf2 effect (Fig. 4E).
Finally, we sought to determine whether Nrf2 can directly bind to the FST promoter using ChIP assays. Data unveiled enrichment of FST promoter region (2.9 fold) but not ACTB gene following immunoprecipitation with the Nrf2 antibody. SiO 2 NP treatment significantly increased Nrf2 binding (5.3 fold) to FST promoter (Fig. 4F). These data indicated that FST gene is a direct target of Nrf2.
FST protects against SiO 2 NP-induced cell apoptosis. SiO 2 NP treatment induces cellular toxicity both in vitro and in vivo 1,18,19 . Since FST expression was up-regulated in response to SiO 2 NP, we suspected that FST has a protective role. To test this hypothesis, the expression of FST was down-regulated using siRNA (Fig. 5A,B) and cell viability was detected. Cell viability significantly decreased in FST silenced cells compared to control cells after SiO 2 NP treatment (Fig. 5C). Decreased cell viability might be due to cell apoptosis. We therefore determined cellular apoptosis by annexin V-FITC staining method. The fractions of annexin V-positive cells were low in control groups (8.7 ± 1.4%, and 6.9 ± 0.9% respectively). SiO 2 NP treatment markedly increased cell apoptosis in control siRNA-transfected cells (18.6 ± 1.4%), and elimination of FST enhanced the apoptotic tendency to 25.3 ± 1.0% (P < 0.01 compared to control) (Fig. 5D,E), indicating FST has a protective role in apoptosis.
FST protects against SiO 2 NP toxicity in vivo. Next, we asked whether FST protects against SiO 2 NP toxicity in lung tissue. To that end, the FST expression was down-regulated using virus-mediated shRNA and Luciferase activity was measured 24 h following plasmid transfection. (F) A549 cells were treated with or without 50 μ g/ml of SiO 2 NPs for 12 h. ChIP analysis was performed with antibodies against Nrf2 and analyzed by qPCR. The values are the mean ± SD of three independent experiments. *P < 0.05, **P < 0.01 (One way ANOVO). mice were treated with SiO 2 NPs for 1 day. Cell apoptosis was then determined by TUNEL assays. Both qPCR and immunoblotting data showed that shRNA significantly reduced FST expression in mouse cells (Fig. 6A,B). Almost no TUNEL-positive cells were observed in the control group, while obvious TUNEL-positive staining was observed in SiO 2 NP-treated lungs, indicating that SiO 2 NP exposure induced cell apoptosis in lung tissue 18,20 . The number of TUNEL-positive cells increased in FST-KD group (Fig. 6C,D), indicating that FST protects against SiO 2 NP toxicity in vivo.

FST inhibits SiO 2 NP-induced ROS production and the expression of NOX1 and NOX5
genes. The toxic effect of SiO 2 NPs is mainly mediated by particle-induced oxidative stress 1 . We therefore determined whether FST exerts its protective function through inhibiting ROS production during SiO 2 NP exposure. SiO 2 NP treatment increased total ROS level in A549 cells. Knockdown of FST further increased ROS level while overexpression of FST decreased ROS level (Fig. 7A,B), indicating an inhibitory role of FST in SiO 2 NP-induced ROS production.
ROS is produced in mitochondria and by cytosolic NADPH oxidases (NOXs) 21,22 . Therefore, FST might inhibit ROS level by regulating either mitochondria or NOX functions. We firstly detected mitochondrial ROS level using MitoSOX assay and data showed that knockdown of FST did not affect mitochondrial ROS production (Fig. 7C). Next, we detected the expression of NOX1, NOX2, NOX3, NOX4, and NOX5, which are involved in cytosolic ROS production 21 . Our data showed that the mRNA levels of NOX2, NOX3 and NOX4 were very low in A549 cells. The mRNA levels of NOX1 and NOX5 significantly increased during SiO 2 NP treatment, indicating that NOX1 and NOX5 are involved in SiO 2 NP-induced cytosolic ROS synthesis. Knockdown of FST increased while overexpression of FST decreased the expression of NOX1 and NOX5 under both control and SiO 2 NP treatment (Fig. 7D-G). These data suggested that FST inhibits NOX1 and NOX5 expression to regulate cytosolic ROS production.

Discussion
The toxicity of SiO 2 NPs has gained much attention with the fast growth of nanotechnology. Both in vitro and in vivo evidence has shown that SiO 2 NPs induce toxic effect in respiratory system 1,18,19 . However, our understanding towards cellular responses to overcome the SiO 2 NPs toxicity is still limited. In this report, we provided experimental evidence to support that FST is involved in cellular protective response to SiO 2 NPs. The major findings of our study include: 1) SiO 2 NPs induced endogenous expression of FST both in mouse lungs and in cultured lung epithelial cells; 2) SiO 2 NPs activated FST transcription through Nrf2; 3) FST inhibited SiO 2 NP-induced apoptosis; 4) FST inhibited ROS production triggered by SiO 2 NPs. These findings demonstrated a protective role of FST (Fig. 8) and might provide a better understanding of the interaction between SiO 2 NPs and biological systems.
FST is a stress responsive gene and its expression can be induced by SiO 2 NP treatment as well as glucose deprivation. However, the mechanism is totally different. We have shown that the induction of FST by glucose deprivation was due to an increase in the half-life of its mRNA. Under normal condition, AUF1 bound to AU-rich element in the 3'UTR of FST mRNA and promoted FST mRNA decay. During glucose deprivation, a majority of AUF1 shuttled from cytoplasm to nucleus, resulting in dissociation of AUF1 from FST mRNA and thus stabilization of FST mRNA 14 . However, our present data showed that the half-life of FST mRNA did not change during SiO 2 NP-treatment. On the contrary, the Ac-H3(K9/18) level and H3K4me2 level at FST promoter region significantly increased, indicating an activation of FST transcription. Therefore, SiO 2 NP regulates FST expression at transcriptional level while glucose deprivation regulates FST expression at post-transcriptional level. This might be due to the different signaling activated. Glucose deprivation activates AMPK pathway and AMPK activation mediates FST induction during glucose deprivation as AMPK inhibition blocked the increase of FST mRNA 14 . On the contrary, SiO 2 NPs mainly induce ROS generation and oxidative stress. SiO 2 NP-induced ROS production activates transcription factor Nrf2, which induces FST transcription. This was supported by the decreased FST expression in Nrf2-deficient cells and the direct binding of Nrf2 to FST promoter region (Fig. 4).
How does FST exert its protective function during SiO 2 NP treatment? NP-induced ROS generation and oxidative stress is the best-developed paradigm to explain the cellular responses to nanoparticles. Our data demonstrated that FST inhibited NOX1 and NOX5 expression and ROS production, indicating that the protective effect of FST is due to the ROS-inhibition. Although the detailed mechanism by which FST inhibits ROS production is still unclear, there are clues pointing to the TGF-β -like molecule-neutralizing effect of FST. It has been reported that TGF-beta family members increased cellular ROS level in a variety of cell lines while FST antagonized their effects. For example, myostatin induced oxidative stress by producing ROS in skeletal muscle   23 . Activin induced endothelial cell oxidative stress and endothelial cell dysfunction, and these effects are mitigated by follistatin 15 . BMP4 treatment increased ROS levels in basal progenitor cells, while follistatin inhibited BMP4-induced ROS production 16 . All these data support that FST is a negative regulator of cellular ROS production.
Nanoparticles were reported to induce general toxic effects on cells such as oxidative stress, inflammation and cell death. However, different NPs have different physic-chemical properties, they might induce NP-specific cytotoxicity. For example, a panel of industrially most relevant metal oxide nanoparticles was screened for toxic effects in A549 cells. Data showed that SiO 2 NPs induced acute cytotoxicity, which is consistent with our results. SiO 2 NPs also induced the expression of COX-2, TNF-α , IL-1β , IL-6 and IL-8. However, iron oxide-or titania-NPs were not cytotoxic and only specifically induced TNF-α expression 24 . Due to the different cytotoxicity, the following cellular protective responses might also be different. Shi J et al. reported that microsomal glutathione transferase 1 (MGST1) protected against SiO2 nanoparticle-induced oxidative stress and cytotoxicity, while it failed to prevent ZnO nanoparticle-triggered cell death 25 . Our data showed that FST can also be induced by TiO 2 NPs, Ag NPs, and ZnO NPs, but the expression level of FST did not change significantly under gold nanorods (GNRs) or graphene oxide (GO) treatment (Fig. S4). These data indicated the existence of different mechanism of cytotoxicity of different nanoparticles.
In summary, we have identified FST as a SiO 2 NP-responsive protein that protects against SiO 2 NP lung toxicity. These findings point to a protective strategy for respiratory system to overcome SiO 2 NP-induced adverse effects.

Materials and Methods
Silica nanoparticles. SiO 2 nanoparticle was from Sigma-Aldrich (#637238, Sigma-Aldrich, St. Louis, MO, United States). Nanoparticles were dispersed in double distilled water at the concentration of 10 mg/mL. Nanoparticles were sonicated at 200 W for 30 seconds prior to cell or mouse treatment.
Transmission electron microscope (TEM) was taken by a JEOL JEM-1200EX transmission electron microscope for nanoparticles. SiO 2 NPs were dispersed in Roswell Park Memorial Institute (RPMI)-1640 medium (Invitrogen, Carlsbad, CA, United States) for 0 h and 24 h and then subjected to dynamic light scattering (DLS) and zeta-potential measurements using the instrument Zetasizer Nano ZS-90 (Malvern Instruments, Orsay, France).

Cells culture and nanoparticle treatment.
Lung epithelial cells (A549) were obtained from ATCC and cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA, United States). Cells were maintained at 37 °C in an atmosphere containing 5% CO2 and 100% humidity.
For SiO 2 NP treatment, cells were seeded in 6-well plates. After 24 hours, cells were treated with silica nanoparticles suspended in RPMI-1640 supplemented with 1% fetal bovine serum at different concentration for indicated time and then subjected to further experiments.
Plasmids. Plasmid expressing FST288 was kindly provided by Dr. Henry T. Keutmann (Massachusetts General Hospital).
For the construction of FST promoter plasmid, FST promoter region was amplified by PCR and cloned to vector pGL3-basic between XhoI and HindIII sites. To construct Nrf2-binding site deletion mutant, the 5' and 3' parts of FST promoter region was amplified and ligated using BamHI site.
For the construction of shRNA plasmid targeting FST, lentviral vector pLVX-shRNA1 was used. shRNA oligonucleotides were designed using online protocol www.clontech.com. Oligo sequences were listed in supplemental Table 1. Synthesized DNA oligos were annealed and ligated into the BamHI/EcoRI-digested pLVX-shRNA1 vector.
RNA purification and reverse transcription reaction. Total RNA was isolated with Trizol reagent (Invitrogen) following the manufacturer's protocol. 0.5 μ g of total RNA was reverse transcribed using random hexamers and the High Capacity cDNA Reverse Transcription Kit (Life Technologies, Grand Island, NY, United States).

Real-time quantitative PCR analysis.
Real-time quantitative PCR analysis was performed in 10-μ l reactions using SYBR GREEN PCR Master Mix (Applied Biosystems). The related mRNA level was normalized to the β-actin mRNA level. Data were analyzed using the 2 −ΔΔCt method 26 . Sequences of all the primers used for PCR amplification are listed in supplemental Table 1.
Immunoblotting analysis. Proteins were quantified by BCA protein assay kit (Bio-Rad) and applied to immunoblotting analysis as described previously [2]. 50 μ g of total proteins were subjected to sodium dodecylsulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membrane (Whatman, Clifton, NJ, United States). Membrane was blocked with 3% bovine serum albumin (BSA) TBS-T buffer (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.05% Tween-20), probed with antibodies targeting to FST (Santa Cruz Biotechnology, Santa Cruz, CA, United States), or ACTB (Cell Signaling Technology, Beverly, MA, United States). Membrane was incubated with horseradish-conjugated secondary antibodies, detected with the SuperSignal West Pico chemiluminescence substrate (Thermo Fisher Scientific), and finally exposed to an X-ray film. Alternatively, membrane was incubated with fluorescent secondary antibodies, and visualized by an Odyssey imaging system (Li-COR Biosciences, Lincoln, NE, United States).
Immunohistochemistry. Embedded mouse lung tissues were deparaffinized with xylene, rehydrated in ethanol, and boiled in 10 mM citrate buffer (pH 6.0) for 30 min for antigen retrieval. Endogenous peroxidase Scientific RepoRts | 6:21133 | DOI: 10.1038/srep21133 was blocked by treatment with 3% H 2 O 2 . After blocking in goat serum for 30 min at room temperature, tissues were incubated with anti-FST (Santa Cruz Biotechnology) at 4 °C overnight. The slides were then visualized with Envision System (DAKO Corporation, Carpinteria, CA, United States) and counterstained with hematoxylin. Images were captured with a wide field microscope (Nikon Eclipse Ti, Nikon Instruments Inc., Shanghai, China).
Luciferase assay. A549 cells were transfected with FST promoter and pRL-TK was co-transfected as the internal control. 24 hours after transfection, cells were treated with SiO 2 NPs for another 12 hours. Cells were then lysed and luciferase activity was measured by the Dual-Luciferase assay system (Promega, Madison, WI, United States). The firefly luciferase activity was normalized to renilla luciferase. ROS detection. Total ROS was detected by using a fluorescent probe 2' ,7'-dichlorofluorescin-diacetate (DCFH-DA) (Beyotime, Nanjing, China). Briefly, cultured A549 cells were treated with or without SiO 2 NPs for 12 h and then incubated with 10 μ M of DCFH-DA at 37 °C for 30 minutes. After rinsing with PBS to eliminate excess DCFH-DA, cells were harvested in 0.5 ml PBS and the fluorescence intensity was monitored with flow cytometry (FCM).
MitoSOX (Invitrogen) was used to detect mitochondrial ROS. Cultured A549 cells were collected and incubated with 10 μ M of MitoSOX at 37 °C for 30 minutes and the fluorescence intensity was monitored with flow cytometry.
Apoptosis assay. Cells seeded in 6-well plates were transfected with siRNAs or plasmids for 48 hours. After CCK8-based viability assay. The effect of SiO 2 NP on cell viability was assessed by Cell Counting Kit-8 (CCK8) assay (Dojindo Laboratory, Japan). A549 cells plated in 96-well plate were transfected with siRNAs for 48 hours. Cells were then treated with SiO 2 NPs for another 12 hours. 10 μ l of CCK8 reagent was added to each well and the cells were incubated for 2 hours at 37 °C. The optical density (OD) at 450 nm was measured by using VarioskanFlash (Thermo Scientific).
shRNA plasmid construction. To produce lentivirus expressing FST-shRNA, the plasmid pLVX-shFST, the envelope plasmid pMD2G and the packaging plasmid psPAX2 were co-transfected into 293T cells. Lentiviral supernatants were harvested through a 0.45 μ m filter 48 hours after transfection. Equal amount of Lentiviral supernatants were used for knockdown of FST in cultured cells and in mouse lung tissue.
Mouse treatment. Male ICR mice of six weeks of age were used for the study. Mice were instilled intratracheally with PBS or 100 μ g of SiO 2 NPs for indicated time and then sacrificed 18 . Mice lung tissue was fixed in 10% formalin or stored in liquid nitrogen for mRNA and protein extraction. All studies were performed following the ethical guidelines of the Ethics Committee of Laboratory Animal Care and Welfare, Zhejiang University, School of Medicine. All studies were approved by the Ethics Committee of Laboratory Animal Care and Welfare, Zhejiang University, School of Medicine.
Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay. TUNEL assays were performed using the TUNEL assay kit (Promega, Madison, WI, United States) to assess cell death following the manufacture's instruction. Localized green fluorescence of apoptotic tissue was detected with the Nikon A1R confocal microscope (Nikon Instruments Inc., Shanghai, China).

Chromatin immunoprecipitation (ChIP).
ChIP assays were performed using the ChIP assay kit (Thermo Fisher Scientific) following the manufacturer's protocol. Briefly, A549 cells or mouse lung tissues were cross-linked with 1% formaldehyde for 10 min at 37 °C. Cross-linking was stopped with 0.125 M glycine. For tissue ChIP, mouse lung tissue was grounded and the cells were resuspended in lysis buffer. After sonication to yield DNA fragments of 300-1000 base pairs, the lysates were cleared by centrifugation, diluted 6-fold with ChIP dilution buffer, and precleared with salmon sperm DNA/protein A-agarose at 4 °C for 1 h. For each immunoprecipitation assay, the lysates were incubated with 2 μ g of anti-K4-dimethylated histone H3 (H3K4me2) (Thermo Fisher Scientific), anti-acetylated histone H3 (Lys9/18) (Ac-H3, K9/18) (Thermo Fisher Scientific), Nrf2 (Abcam, Cambridge, MA, United States) or control IgGs (Santa Cruz Biotechnology) overnight at 4 °C with rotation. The immunecomplexes were then collected with protein A-agarose slurry, eluted, and de-crosslinked at 65 °C. After RNase digestion and proteinase digestion, immunoprecipitated DNA was extracted. The purified DNA was amplified by real time PCR with SYBR GREEN PCR Master Mix (Applied Biosystems).
Sequences of the primers used for FST promoter amplification are listed in supplemental Table 1. Statistical analysis. The experiments were repeated at least three times and data were presented as mean ± SD. Statistical significance between two groups was determined with the Student's t-test. Statistical significance among more than 3 groups was determined with one-way analysis of variance (ANOVA). All comparisons were two-tailed and P < 0.05 was considered significant.