Downregulated brain and muscle aryl hydrocarbon receptor nuclear translocator-like protein-1 inhibits osteogenesis of BMSCs through p53 in type 2 diabetes mellitus

ABSTRACT The bone marrow mesenchymal stem cells (BMSCs)-mediated abnormal bone metabolism can delay and impair the bone remodeling process in type 2 diabetes mellitus (T2DM). Our previous study demonstrated that the downregulation of brain and muscle aryl hydrocarbon receptor nuclear translocator-like protein 1 (BMAL1), a circadian clock protein, inhibited the Wnt/β-catenin pathway via enhanced GSK-3β in diabetic BMSCs. In this article, we confirmed that the downregulated BMAL1 in T2DM played an inhibitory role in osteogenic differentiation of BMSCs. Upregulation of BMAL1 in the diabetic BMSCs significantly recovered the expression pattern of osteogenic marker genes and alkaline phosphatase (Alp) activity. We also observed an activation of the p53 signaling pathways, exhibited by increased p53 and p21 in diabetic BMSCs. Downregulation of p53 resulting from overexpression of BMAL1 was detected, and when we applied p53 gene silencing (shRNA) and the p53 inhibitor, pifithrin-α (PFT-α), the impaired osteogenic differentiation ability of diabetic BMSCs was greatly restored. However, there was no change in the level of expression of BMAL1. Taken together, our results first revealed that BMAL1 regulated osteogenesis of BMSCs through p53 in T2DM, providing a novel direction for further exploration of the mechanism underlying osteoporosis in diabetes.


INTRODUCTION
As a chronic metabolic disorder disease, diabetes mellitus (DM) has affected millions of people, and its prevalence has been increasing significantly worldwide (Guariguata et al., 2014). More than 90% of these patients are diagnosed with type 2 diabetes mellitus (T2DM) (McGurnaghan et al., 2019). Most of them are suffering from osteoporosis and following fractures. Bone marrow mesenchymal stem cells (BMSCs) are vital for bone regeneration, due to its multidirectional differentiation potential and the differentiation ability into osteoblasts. Previous studies revealed that the proliferation and multipotency of BMSCs might change in diabetic pathological microenvironment (Zhou et al., 2016). BMSCs-mediated imbalance of bone formation and bone resorption would occur in T2DM (Brown et al., 2014). However, there have been few articles on the molecular mechanisms of diabetic osteoporosis until now.
The core cellular circadian pacemaker in mammals drives circadian rhythms of behaviors, which is coupled to the environmental light cycle (Honma, 2018). Brain and muscle aryl hydrocarbon receptor nuclear translocator-like protein 1 (BMAL1) is a core component of the circadian clock, which has been reported to express universally in both suprachiasmatic nucleus and peripheral tissues and regulate various cellular metabolic processes (Zhang et al., 2014). It combines with CLOCK to form heterodimers and activates transcriptional activity of downstream genes, affecting numerous physiological and biochemical reactions (Gaucher et al., 2018). Researchers have found the involvement of BMAL1 in regulating the rhythmic secretion of pancreatic islet B cells, and its deletion will lead to the occurrence of T2DM (Albrecht, 2017). The expression pattern of BMAL1 was reported to be disrupted in the microenvironment of T2DM (Weger et al., 2017). Moreover, decreased osteogenesis ability of BMSCs was observed in BMAL1 deletion or inhibition mice (Miyamoto et al., 2010). Our previous study preliminarily verified that BMAL1 regulated the osteogenesis of diabetic BMSCs by modulating GSK-3β (Li et al., 2017). However, more work remains to be done to ascertain the mechanism between BMAL1 and osteogenic differentiation of diabetic BMSCs.
Recently, classical tumor suppressor p53 has been reported to regulate glucose metabolism and osteogenic differentiation of BMSCs (Labuschagne et al., 2018). The p53 mutant mice showed symptoms of premature aging, and increased bone density and bone formation ability (He et al., 2015). p53 was illustrated to suppress the expression of runt-related transcription factor 2 (Runx2) and osterix (Osx), which were identified as master transcription factors that control the differentiation of BMSCs into osteoblasts, and modulate the expression of several osteogenic differentiation-related microRNAs (He et al., 2015). Studies have also confirmed that BMAL1 performed as a putative regulator of p53 in cancer (Mullenders et al., 2009). However, the association between BMAL1 and p53 in BMSCs is still uncertain, let alone in T2DM.
In this study, we investigated the inhibitory effect of BMAL1 on osteogenic differentiation of BMSCs in type 2 diabetic rat models. And, with the help of p53 gene silencing (shRNA) and the p53 inhibitor, pifithrin-α (PFT-α), we identified the key molecule p53, which could mediate the regulatory function of BMAL1 on osteogenic differentiation of BMSCs. Our results indicate that the decreased expression of BMAL1 inhibits osteogenesis of BMSCs through p53 in T2DM.

Characteristics of BMSCs
Flow cytometric analysis of phenotypic expression showed that CD34 and CD45 were negatively expressed, while CD29, CD44 and CD90 were positively expressed in cells from both groups (Fig. 1A). These cells isolated from rats' femurs were able to differentiate into osteoblasts or adipocytes after induction of osteogenic or adipogenic differentiation, as demonstrated by Alizarin Red staining and Oil Red O staining (Fig. 1B). The results conformed with previous reports (Degen et al., 2016;Kang et al., 2015). Therefore, cells separated from Wistar rats and Goto-Kakizaki (GK) rats showed the typical characteristics of BMSCs.

Overexpression of BMAL1 rescued the impaired osteogenic differentiation of BMSCs in T2DM
Our previous research has demonstrated that expression of BMAL1 was clearly downregulated in diabetic BMSCs (Li et al., 2017). However, the role of BMAL1 in the regulation of osteogenesis should be assayed in a more rigorous and well-structured way. In this study, we reconfirmed that BMAL1 expression was evidently decreased in diabetic BMSCs, after stable overexpression of BMAL1 in diabetic GK BMSCs with lentiviral infection. When BMAL1 expression was significantly upregulated at both mRNA and protein levels, successful osteogenic differentiation was indicated by the upregulation of the osteogenesis related markers Runx2, Osx and alkaline phosphatase (Alp) ( Fig. 2A,B). But, the expression level of osteocalcin (Ocn), a marker of terminal osteogenic differentiation, showed no statistical difference between diabetic GK BMSCs (GK) and BMAL1 overexpressed diabetic GK BMSCs (GK-BMAL1) (Fig. 2B). What is more, BMAL1 overexpression led to a significant upregulation of osteogenesis related markers Runx2, Alp and Osx at both mRNA and protein levels, after osteogenic differentiation for 7 days ( Fig. 2A,B). After 7 days of osteogenic induction, the Alp activity was significantly lower in diabetic GK BMSCs (GK), but higher in BMAL1 overexpressed diabetic GK BMSCs (GK-BMAL1), compared with the control wild-type (WT) Wistar BMSCs (Fig. 2C). Based on the data above, it is concluded that BMAL1 has an important effect on the regulation of BMSCs osteogenic differentiation in T2DM. BMAL1 overexpression helps rescue the impaired osteogenic differentiation ability of diabetic GK BMSCs.

Decreased BMAL1 level inhibited proliferation of BMSCs in T2DM
Sufficient cells are the basis of normal differentiation of BMSCs, so it is plausible that the proliferation and differentiation of BMSCs are correlative (Liu and Li, 2010). Results of CCK-8 assay revealed that diabetic GK BMSCs had a decreased cell growth rate compared with WT Wistar rats. Overexpressing BMAL1 in GK BMSCs helped recover its proliferation, and its proliferation rate was even higher than that of WT group (Fig. 3A). Cell apoptosis was determined by the Annexin V-FITC/PI apoptosis detection kit. As shown in Fig. 3B, overexpression of BMAL1 by lentiviral infection in diabetic GK BMSCs greatly reduced the percentage of apoptotic cells. Similarly, according to the Alp staining results, GK-BMAL1 BMSCs showed the highest Alp activity, while GK BMSCs had the lowest Alp activity (Fig. 3C). Even up to passage 6, when Alp activity had an age-dependent decrease during osteogenesis, the osteogenic differentiation ability of BMSCs in GK-BMAL1 group was still the highest among the three groups (Fig. 3C). These results indicate that osteogenic differentiation potential of BMSCs in T2DM is affected by BMAL1 expression pattern, which is also related to the proliferation of BMSCs.

Downregulated BMAL1 promoted p53 expression in T2DM
Previously, BMAL1 was reported to perform as a putative regulator of p53 in pancreatic cancer (Jiang et al., 2016). Recently, p53 has been reported to inhibit osteogenic differentiation of BMSCs by suppressing the expression of several osteogenic transcription factors, such as Runx2 and Osx (Oren, 2019). Therefore, we wondered whether there was a correlation between BMAL1 and p53 in the regulation of osteogenic differentiation of diabetic BMSCs.
Our results validated that the expression patterns of p53 and p21, two p53 signaling pathway-related proteins, were significantly upregulated at the protein level in diabetic GK BMSCs (Fig. 4A). In addition, after BMAL1 was overexpressed in GK BMSCs, the expressions of p53 and p21 were clearly decreased (Fig. 4A). Same changes were observed at mRNA level ( Fig. 4B). At the same time, the downregulation of p53 expression level resulted from BMAL1 upregulation was also confirmed by immunofluorescence (Fig. 4C). Results above enlighten us that the process of BMAL1 regulating osteogenic differentiation of diabetic BMSCs may be mediated by p53.

Inhibition of osteogenic differentiation of BMSCs by BMAL1 downregulation in T2DM was mediated by p53
To investigate whether the recovery of osteogenic differentiation capability of BMAL1 overexpressed diabetic BMSCs was mediated by p53, pifithrin-α (PFT-α), a well-known p53 inhibitor, was used in the osteogenic differentiation process of GK BMSCs and GK-BMAL1 BMSCs. Treatment with PFT-α in both GK BMSCs and GK-BMAL1 BMSCs resulted in an increased expression of osteogenesis related markers Runx2, Osx and Alp at both protein and mRNA levels ( Fig. 5A,B). The GK-BMAL1 BMSCs with PFT-α showed a further upregulation of osteogenesis markers than GK-BMAL1 BMSCs (Fig. 5A,B). Meanwhile, Alp activity was largely recovered in GK and GK-BMAL1 BMSCs after treatment with PFT-α (Fig. 5C). Besides, no statistically significant changes in BMAL1 expression level were observed at mRNA or protein level (Fig. 5A,B). However, reports reminded us of the potential off-target effects of PFT-α (Kanno et al., 2015). In this regard, the same conclusions above were reconfirmed by p53 gene silencing (shRNA). As shown in Fig. 6A-C, upregulated osteogenic markers (Runx2, Osx and Alp) and higher Alp activity were observed in GK BMSCs and GK-BMAL1 BMSCs, when p53 gene expression was silenced. Application of p53 gene silencing also had no influence on the BMAL1 expression pattern (Fig. 6A-C). In summary, the relationship between BMAL1 and p53 was effectively demonstrated. Impaired osteogenic differentiation ability of diabetic GK BMSCs could be partly rescued by p53 shutdown. Interferences of p53 expression had no effect on BMAL1 expression level at neither mRNA nor protein level. Collectively, these data suggest that the role of BMAL1 in osteogenic differentiation of BMSCs in T2DM is mediated by p53.

DISCUSSION
Long-term hyperglycemia will lead to the disruption of bone homeostasis, resulting in impaired bone formation and increased  bone resorption (Karner and Long, 2018). Research has shown that bone metabolic disorders are primarily caused by the defect in bone formation in T2DM (Farr and Khosla, 2016). At present, there are few studies on the molecular mechanism of osteoporosis in diabetes. Bone homeostasis is mediated by mesenchymal stem cell-derived osteoblasts and hematopoietic-derived osteoclasts, respectively (Raggatt and Partridge, 2010). BMSCs are multipotent progenitors capable of forming multiple tissue types and central mediators of osteogenesis (Bianco et al., 2008). The multipotential differentiation ability of BMSCs was suppressed in T2DM (Aikawa et al., 2016). Therefore, BMSCs could be the ideal target for researches of osteoporosis in T2DM.
It was in the 1960s that circadian clock components such as BMAL1 were first discovered to impair the mitochondrial function in pancreatic B cells (Jarrett and Keen, 1969). When circadian clock components such as BMAL1 are disrupted, hypoinsulinism and diabetes occur (Marcheva et al., 2010). The BMAL1-knockout mice showed a lack of rhythm in insulin action and disrupted insulin responsiveness, which were characteristics of T2DM (Sadacca et al., 2011). Decreased BMSC osteogenesis ability was also observed in BMAL1 inhibition mice (Kondratov et al., 2006). The correlation between BMAL1 and osteogenic differentiation of BMSCs in T2DM could be an interesting area. Previously, we have reported the mechanism underlying the regulation of BMAL1 on osteogenesis in a preliminary way. We verified that decreased BMAL1 upregulated GSK-3β and inhibited Wnt/β-catenin pathway in type 2 diabetic rat models, leading to the decreased osteogenic ability of diabetic GK BMSCs (Li et al., 2017). However, it may contain several flaws in reasoning. Here, our results demonstrated again in a relatively scientific way that upregulated BMAL1 restored osteogenic differentiation of BMSCs in T2DM (Fig. 2).
It is commonly known that the osteogenic differentiation of stem cells is controlled by a variety of lineage-specific transcription factors, including Runx2 and Osx (Liu et al., 2016). Several noncoding RNAs and proteins have been reported to regulate the expression of these transcription factors (Farr and Khosla, 2019). Our results demonstrated that upregulated BMAL1 expression influenced the expression of Runx2, Osx and Alp (Fig. 2). Analysis of data showed that BMAL1 expression had no significant effect on the expression level of Ocn ( Fig. 2A). It is understandable that Ocn is a marker of terminal stages osteogenic differentiation, which signifies the conversion of osteoblast to osteocyte and always expresses after 21 days of osteogenic differentiation induction in vitro (Tataria et al., 2006). These results were the same as the previous report that BMAL1 −/− mice had a low bone mass phenotype (Samsa et al., 2016).
Stem cells would eventually differentiate into specific cells, and we exampled the results of osteogenic differentiation with Alp activity staining in this study. The differentiation of stem cells must involve differences between cell proliferation and apoptosis. According to our results, the variations of BMAL1 expression did change the states of BMSCs' proliferation and apoptosis (Fig. 3A,B). The effect that upregulated BMAL1 promoted the osteogenic differentiation of BMSCs existed continuously until the sixth passage (Fig. 3). Some mechanisms could be explored among this phenomenon other than GSK-3β pathway, which we have published about previously (Li et al., 2017). In T2DM, p53 modulates blood glucose by interfering with glycolysis, oxidative phosphorylation and pentose phosphate (Halim et al., 2019). The p53 signaling pathways have been widely reported due to its negative regulation of bone formation (Kastenhuber and Lowe, 2017). Moreover, p53 has also been substantiated to suppress the expression of Runx2 and Osx in vitro differentiation models (Tataria et al., 2006). We wondered whether BMAL1 could regulate osteogenic differentiation of BMSCs by modulating the p53 expression in T2DM. As shown in Fig. 4, the upregulated p53 expression was observed in diabetic GK BMSCs, while the expression of p53 significantly decreased at both protein and mRNA levels after BMAL1 overexpression using lentiviral infection in diabetic GK BMSCs. The conclusions above were also confirmed by immunofluorescence (Fig. 4C). With the treatment of PFT-α and p53 gene silencing, decreased p53 level and upregulated expressions of osteogenic markers were detected (Figs 5 and 6). Alp staining analysis verified the same results as above (Figs 5 and 6). Moreover, inhibition of p53 expression pattern could not reduce the BMAL1 expression, which was significantly increased by lentiviral infection (Figs 5 and 6). All these results demonstrate that downregulated BMAL1 inhibits the osteogenic differentiation potential of BMSCs in T2DM, in a partially p53-dependent manner. To our knowledge, this is the first report of the relationship between BMAL1 and p53 in the regulation of osteogenic differentiation of BMSCs in T2DM.
However, what we have done still leaves much to be desired. Firstly, although we want to examine the relationship between BMAL1 and p53 in T2DM perfectly, we could not find a relatively ordinary T2DM control cell line, which could be used to perform experiments about knocking-down BMAL1 expression to observe the change of p53. The particularity of the pathological environment of T2DM always results in a significantly reduced BMAL1 expression (Marcheva et al., 2010). As for WT Wistar rats, they cannot simulate the microenvironment of T2DM, and we do not know whether upregulation of BMAL1 in their BMSCs would make the cells abnormal, such as the occurrence of biorhythm disorder. Therefore, in this study, we only referred BMSCs of WT Wistar rats as a datum dimension. Secondly, considering the particularity of p53, deeper research need to be done about the specific molecular mechanisms between BMAL1 and p53. Reports have already shown that the evolutionarily conserved p53 response element overlaps with the E-BOX element critical for BMAL1/ CLOCK binding, which suggests a specific location of interaction between p53 and BMAL1 (Miki et al., 2013). Thirdly, we noticed the p53 shutdown could not fully restore the impaired osteogenesis as the BMAL1 overexpression by lentiviral infection did (Figs 5 and 6). There must be some other molecules involved in the regulation of osteogenesis of BMSCs by BMAL1. In conclusion, our research reveals for the first time that downregulated BMAL1 inhibits osteogenesis of BMSCs through p53 in T2DM. The results presented in our study suggest a potential therapeutic strategy in clinical practice to treat osteoporosis in T2DM.

Animal care
Ten 8-week-old male GK rats, a well-known spontaneous model of T2DM (D'Souza et al., 2011), and ten age-sex-matched Wistar rats, which have the same genetic background as GK rats, were obtained from Kavens Experimental Animals Co., Ltd. (Changzhou, China). All rats were kept in the Chinese PLA General Hospital Research Institute mouse facility at 22°C on a 12 h light-dark cycle with free access to water and food for 1 month. All experiments were conducted in accordance with the guidelines laid down by the National Institute of Health (NIH), USA, and approved by the Institutional Animal Care and Use Committee of Chinese PLA General Hospital.

Cell culture
After regular detection of body weight, fasting blood glucose level, fasting insulin level and oral glucose tolerance tests (OGTT), eight 12-week-old male stable T2DM GK rats were selected. While ten age-sex-matched Wistar rats were used as the control group. Then all rats were euthanized via intraperitoneal injection of excess pentobarbital at the same time (100 mg kg −1 ). Under aseptic conditions, we isolated BMSCs from femurs by cutting the ends of them and flushing the marrow with 5 ml α-minimum essential medium (α-MEM; Hyclone, UT, USA) using a syringe fitted with a 25-gauge needle. All cells were cultured in α-MEM supplemented with 10% fetal bovine serum (FBS; GIBCO, NY, USA) and 100 U ml −1 penicillinstreptomycin (Hyclone) at 37°C with 5% CO 2 in a completely humidified atmosphere. The medium was replaced every 3 days. Cells were detached using 0.25% trypsin and ethylenediaminetetraacetic acid (Trypsin-EDTA; Hyclone) when they reached 80-90% confluence. All BMSCs used in this study were passage 3 BMSCs except special situations with corresponding descriptions.

Identification of BMSCs
BMSCs from Wistar rats and GK rats were incubated with fluorescein isothiocyanate (FITC)-conjugated monoclonal antibodies for CD29, CD34, CD44, CD45 and CD90 (Abcam, Cambridge, UK) to investigate the cluster of differentiation (CD) phenotype of BMSCs. All cells were subjected to flow cytometric analysis using a BD FACS Calibur flow cytometry (BD Biosciences, NJ, USA).

Induction of osteogenic differentiation
Before the medium was replaced with osteogenic induction medium, BMSCs were cultured in the basal medium until they reached approximately 80% confluence. To induce osteogenic differentiation, BMSCs were cultured in osteogenic induction medium with the following compositions: α-MEM supplemented with 10% FBS, 100 U ml −1 penicillin-streptomycin, 1 mM dexamethasone, 1 M β-glycerophosphate and 10 mM ascorbic acid (Solarbio, Beijing, China). The osteogenic induction medium was replaced every 3 days. Alp activity assay of the differentiated cells was conducted with a BCIP/NBT Alkaline Phosphatase Color Development Kit (Beyotime, Shanghai, China) according to the manufacturer's instructions. Alizarin Red staining was used to assay the deposition of calcium into the extracellular matrix according to standard protocols.

Induction of adipogenic differentiation
BMSCs from each group were maintained in six-well plates at a density of 5×10 5 cells well −1 . After they reached about 80% confluence, we replaced the basal medium with adipogenic induction medium, composed of α-MEM supplemented with 0.5 mM methylisobutylxanthine, 0.5 mM hydrocortisone, 60 mM indomethacin, 10% FBS, and 100 U ml −1 penicillin-streptomycin. The adipogenic induction medium was replaced every 3 days. All BMSCs were incubated for 3 weeks in adipogenic induction medium, and then Oil Red O staining was performed to assay intracellular lipid accumulation according to standard protocols.

Infection of BMAL1 overexpression lentiviral vector
Diabetic GK BMSCs were cultured in six-well plates, and lentivirus infection was performed after cells reached 80% confluence. BMAL1 overexpression lentiviral vector was purchased from Inovogen Tech. Co., Ltd. (Beijing, China; Gene ID: 29657) and diabetic GK BMSCs were infected with lentivirus at a multiplicity of infection of 20 according to the corresponding manufacturer's instructions (Inovogen Tech. Co., Ltd., Beijing, China). Diabetic GK BMSCs in the experimental group (GK-BMAL1 BMSCs) were infected with BMAL1 overexpression lentiviral vector, BMSCs in positive control group were infected using enhanced green fluorescent protein (EGFP)-expressing lentiviral vector, and BMSCs without any infection were defined as negative control group.
Cell counting kit-8 (CCK-8) assay BMSCs were seeded in 96-well plates at a density of 5×10 4 cells well −1 and then cultured in the incubator at 37°C with 5% CO 2 . 10 µl CCK-8 solution (Solarbio, Beijing, China) was added to each well at the indicated time points, and absorbance at 450 nm was measured by enzyme labeling instrument after 4 h of incubation. The OD 450 value was measured from the first day to the seventh day in each well per group. To plot the growth curve, the cultured time was considered as the abscissa and the OD 450 value was considered as the ordinate.

Flow cytometry analysis of cell apoptosis
BMSCs from WT Wistar group, diabetic GK group and GK-BMAL1 group were cultured for 5 days and then the cell apoptosis assay was performed with an Annexin V-FITC/PI Apoptosis Detection Kit (Solarbio, Beijing, China) according to the manufacturer's instructions using a BD FACS Calibur flow cytometry.

Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA of BMSCs was extracted with TrIquick Reagent (Solarbio, Beijing, China). The concentration of total RNA was quantified using a spectrophotometer at the absorbance of 260 nm and 280 nm. A260/A280 ratios of all the samples were between 1.9 to 2.0, indicating high purity. Total RNA from each sample was subjected to reverse transcription to cDNA with a PrimeScript™ RT Reagent Kit (TaKaRa, Japan). We conducted qRT-PCR on a Bio-Rad CFX96 Touch (Bio-Rad, Hercules, California, USA) with 20 µl qRT-PCR mix, which contains 10 µl TransStart Top Green qPCR SuperMix (TransGen Biotech, Beijing, China), 0.8 µM of each primer, 1 µl passive reference Dye-II, 1 µl cDNA and distilled water (containing diethylpyrocarbonate). The thermal cycling conditions were 95°C for 30 s, followed by 45 cycles of 95°C for 5 s, 60°C for 30 s, and increments of 0.5°C for 5 s from 65°C to 95°C for the melting curve. β-actin was used as an internal control. Primer sequences used for qRT-PCR are listed in Table S1.

Immunofluorescence
BMSCs were fixed with 4% paraformaldehyde for 20 min, permeabilized with 0.1% Triton X-100 for 5 min, and then blocked in 5% BSA for 1 h. Samples were incubated with a p53 polyclonal antibody (#ab26, Abcam) at a dilution of 1:100 at 4°C overnight. Subsequently, they were washed for three times with phosphate buffered solution (PBS; Solarbio, Beijing, China) and incubated with the corresponding secondary antibody in 5% BSA for 1 h. Then the nuclei were labeled with 1 mg ml −1 DAPI for 15 min. Cells were visualized by a confocal laser scanning microscope after they were washed with PBS for three times.

Inhibition and knockdown of p53
Inhibition of p53 was conducted by adding the famous p53 inhibitor pifithrin-α (PFT-α; Selleck, Houston, TX, USA) (20 μM) to the osteogenic induction medium described above and renewing the medium every 3 days. The lentiviral vector carrying specific p53-targeting shRNA was obtained from GenePharma Co., Ltd. (Shanghai, China; Gene ID: 24842). BMSCs were infected with the lentiviral vector at a multiplicity of infection of 100 according to the protocols described above. The sequence of p53 shRNA was 5′-GAG AAT ATT TCA CCC TTA A-3′ .

Statistical analysis
Data were expressed as mean±s.e.m. and analyzed with SPSS 19.0 software (SPSS Inc., IBM, Chicago, USA). All experiments were performed at least in triplicate. According to the study design, an unpaired Student's t-test or ANOVA with post hoc test was used to determine the statistical significance. The threshold for statistical significance was set at *P<0.05; # P<0.03; **P<0.01.