MicroRNA-145 attenuates TNF-α-driven cartilage matrix degradation in osteoarthritis via direct suppression of MKK4

Cartilage dyshomeostasis contributes to osteoarthritis (OA) pathogenesis, and tumor necrosis factor (TNF)-α has critical role in this process by driving inflammatory cascades and cartilage degradation. However, the negative regulation of TNF-α-mediated signaling remains undefined. Here we demonstrate the crucial role of miR-145 in the modulation of TNF-α-mediated signaling and cartilage matrix degradation. MicroRNA (miRNA) expression profiles of TNF-α-stimulated chondrocytes showed that miR-145 expression was rapidly downregulated by TNF-α. Moreover, miR-145 was directly repressed by p65 and was negatively correlated with TNF-α secretion during OA progression. Further, we found that miR-145 directly targeted mitogen-activated protein kinase kinase 4 (MKK4) and broadly restrained the production of several TNF-α-triggered matrix-degrading enzymes (MMP-3, MMP-13, and Adamts-5). Mechanistic studies unveiled that miR-145 negatively regulated TNF-α-mediated JNK and p38 activation, as well as the nuclear accumulation of p-c-Jun and p-ATF2, by inhibiting MKK4 phosphorylation, eventually resulting in the alteration of catabolic genes transcription. Indeed, p-ATF2 interacted with the promoter of Mmp-13, whereas p-c-Jun bound to promoters of Mmp-3 and Adamts-5. MKK4 was significantly elevated in OA cartilage. Eliminating MKK4 by short hairpin RNA resulted in obviously decreased matrix-degrading enzymes production, JNK and p38 inactivation, and an inhibition of cartilage degradation. On the contrary, MKK4 overexpression enhanced TNF-α-mediated signaling activation and transcription of downstream catabolic genes, and consequently worsened cartilage degradation. Moreover, intra-articular (IA) injection of miR-145 agonist to rat with surgery-induced OA alleviated cartilage destruction. Altogether, we elucidate a novel regulatory mechanism underlying TNF-α-triggered cartilage degradation and demonstrate the potential utility of miR-145 and MKK4 as therapy targets for OA.

Osteoarthritis (OA) is the most prevalent chronic joint disease characterized by a group of abnormalities, such as cartilage destruction, subchondral bone remodeling, and synovial inflammation. 1 These conditions have been mainly attributed to an imbalance between the anabolism and catabolism of the articular cartilage, especially an increase in catabolism. 2 Proinflammatory cytokines, such as interleukin-1β (IL-1β), tumor necrosis factor (TNF-α), and IL-6, are critical mediators in this dyshomeostasis. [3][4][5] In general, cytokines such as IL-1β and TNF-α increase matrix-degrading enzyme (matrix metalloproteinases, MMPs and a disintegrin and metalloproteinase with thrombospondin motifs, ADAMTS) gene expression, which in turn degrade the cartilage extracellular matrix (ECM), and then the degraded ECM fragments further facilitate synovial inflammation and the production of proinflammatory cytokines, and eventually lead to cartilage destruction. 4,6 On the other hand, cytokines also blunt the synthesis of the cartilage ECM. 7 Therefore, molecules possessing anti-inflammatory properties are urgently needed.
It is known that inflammation presents cascade amplification, 8 implying that signaling cascades consisting of protein kinases, especially the upstream molecules involved in NF-κB and mitogen-activated protein kinases (MAPK) signaling pathways, may have vital roles in driving the inflammatory cascade during OA. In addition, as the main cytokines involved in OA pathogenesis, IL-1β is mainly associated with cartilage destruction, whereas TNF-α is implicated in driving of the inflammatory cascades during OA. 4,9 Furthermore, chondrocytes from human OA cartilage present abnormal high expression of the p55 TNF-α receptor, which renders OA cartilage, particularly susceptible to TNF-α degradative stimuli. Meanwhile, OA cartilage produces more TNF-α and TNF angle convertase enzyme compared with normal cartilage. 10 Therefore, targeting TNF-α signaling may be a powerful strategy to cure OA.
Mechanisms that mediate the actions of TNF-α have been intensively studied. 11,12 In general, upon TNF-α binding to TNF receptor (TNFR), a conformational change promotes recruitment of cytosolic factors to the TNFR intracellular death domain (DD). Subsequently, TNFR-associated DD (TRADD) is recruited directly via a homotypic DD-DD interaction, and then the TRADD adaptor recruits a DD-containing kinase termed receptor interacting protein-1 (RIP1). RIP1 in turn recruits TNF receptor associated factor-2 (TRAF2) and/or TRAF5, as well as cellular inhibitor of apoptosis protein-1 (cIAP1) and cIAP2. These adaptors act in concert with RIP1 to engage the downstream phosphorylation cascades to activate NF-κB, JNK, and p38 pathways, which ultimately promote inflammation and transcriptional activation. [13][14][15] However, the negative regulation of TNF-α-mediated signal transduction, especially the modulation of upstream molecules still remains largely unknown.
In this study, we explore TNF-α-responsive miRNAs in chondrocytes and attempt to identify miRNAs intimately implicated in TNF-α-mediated signaling and investigate their potential roles in cartilage matrix degradation during OA pathogenesis. Our results show that miR-145 is sensitive to TNF-α, characterized by a rapid reduction upon TNF-α stimulation. In turn, miR-145 suppresses TNF-α-mediated JNK/c-Jun and p38/ATF2 activation and induction of MMP-3, MMP-13, and Adamts-5 by directly targeting mitogenactivated protein kinase kinase 4 (MKK4), and consequently attenuates cartilage matrix degradation in experimental OA.

Results
MiR-145 is downregulated in TNF-α-stimulated chondrocytes and OA cartilage. We first detected the levels of proinflammatory cytokines during surgery-induced OA and found that IL-1β and TNF-α were strongly enriched compared with other cytokines (Figure 1a). Treatment of chondrocytes with TNF-α promptly triggered the activation of NF-κB and MAPK pathways (Supplementary Figure 1a), along with the nuclear import of p-p65, p-c-Jun, and p-ATF2 (Supplementary Figure 1b), resulting in the upregulation of MMP-3, MMP-13, and Adamts-5, which play crucial roles in OA cartilage destruction ( Supplementary Figures 2a-g). Expectedly, blocking NF-κB or MAPK partially restrained TNF-α-induced upregulation of matrix-degrading enzymes ( Supplementary Figures 1c and d). Similarly, CC-5013 (TNFα secretion inhibitor) reduced C2C (a cleavage neoepitope of collagen II) concentration, cartilage destruction, and the upregulation of MMP-3, MMP-13, and Adamts-5 during experimental OA (Figures 1b and c and Supplementary Figure 3a). Upon TNF-α stimulation, 15 miRNAs were upregulated more than 1.5-fold, whereas 18 miRNAs showed a 435% decrease (Figures 1d and e). Results of qRT-PCR validation showed that the expression of miR-23b and miR-145 were indeed reduced by TNF-α in time-and dose-dependent manners (Figures 1f and g). Of particular interest, miR-145 was selected for further investigation because of its higher expression level in chondrocytes and cartilage compared with miR-23b ( Supplementary Figures 4a and b). In addition, miR-145 was markedly decreased in human OA cartilage, characterized by higher levels of TNF-α, MMP-3, MMP-13, and Adamts-5 and obvious destruction of cartilage matrix, compared with the normal cartilage (Figures 1h and i). Moreover, a solid inverse correlation was observed between miR-145 and TNF-α in surgery-induced OA and human OA cartilage (Figure 1j). Interestingly, CC-5013 significantly restored the reduction of miR-145 caused by destablization of the medial meniscus (DMM) surgery ( Figure 1k). Altogether, these results demonstrate that miR-145 is negatively regulated by TNF-α in both chondrocytes and OA cartilage.
The downregulation of miR-145 by TNF-α is dependent of the tradd-traf2-p65 axis. To elucidate the underlying mechanism of miR-145 reduction by TNF-α, NF-κB and MAPK inhibitors were used. Results showed that BAY11-7082 (NF-κB inhibitor) extremely rescued the inhibitory effects of TNF-α on miR-145 expression ( Figure 2a). We then performed RNA interference (RNAi)-mediated knockdown of p65, traf2, and tradd in chondrocytes (Figures 2b-d). Indeed, knockdown of p65, traf2, or tradd blocked TNF-αinduced downregulation of miR-145 ( Figure 2e). Bioinformatic analyses reveal that there is a consensus NF-κBbinding site in miR-145 promoter ( Figure 2f). Thus, we performed chromatin immunoprecipitation (ChIP) assay and confirmed that TNF-α stimulation led to recruitment of NF-κB subunit, p65, to the promoter of miR-145 ( Figure 2g). In cells expressing a luciferase construct containing the wild-type miR-145 promoter, TNF-α markedly reduced luciferase activity, whereas mutation of the NF-κB-binding site in the promoter inhibited TNF-α-triggered suppression of reporter activity (Figure 2h), indicating that p65 binding to the miR-145 promoter is crucial for the reduction of miR-145 by TNF-α. These results collectively suggest that reduction of miR-145 by TNF-α is mainly through the canonical NF-κB signaling pathway.
MiR-145 directly targets MKK4. Next, we investigated the possible targets of miR-145. Bioinformatic analysis using miRNA target prediction software revealed that MKK4, a member of the MKKs family, was a putative target of miR-145 ( Figure 4a). Notably, miR-145 mimics reduced, whereas miR-145 inhibitor elevated, MKK4 and p-MKK4 in a dosedependent manner under TNF-α stimulation (Figures 4b-d).
Similarly, miR-145 overexpression plasmid inhibited, whereas miR-145 inhibition plasmid promoted, the expression of MKK4 (Figures 4e and f). We further observed that the  Figure 4h). To obtain more direct evidence, luciferase reporter constructs were generated and co-transfected with miR-145 mimics or inhibitors into the SW1353 cell line. Results showed that miR-145 mimics significantly inhibited, whereas miR-145 inhibitors enhanced, the luciferase activity of cells transfected with wild-type MKK4 3′-UTR, but not cells transfected with the mutant MKK4 3′-UTR (Figure 4i). We then assessed MKK4 expression profile in different tissues from rats. Despite the expression level of MKK4 in the brain, liver, and heart were much higher, MKK4 was found to be substantially expressed in articular cartilage compared with other joint tissues (bone, synovium, and tendon), which suggest that MKK4 may have a functional role in cartilage under physiological or pathological conditions ( Figure 4j). Furthermore, MKK4 and its downstream molecules was markedly elevated in OA-affected, damaged regions of human cartilage (Figures 4k and l) and surgeryinduced OA cartilage (Figure 4m), supporting a potential role of MKK4 in OA pathogenesis. Collectively, these results suggest that MKK4 is a direct target of miR-145 and that the endogenous MKK4 is tightly regulated by miR-145 in chondrocytes during TNF-α stimulation and OA pathogenesis.

MiR-145 negatively regulates TNF-α-mediated signaling.
To elucidate the underlying mechanism through which miR-145 inhibits the production of TNF-α-induced matrixdegrading enzymes, we assessed the effect of miR-145 on the activation of NF-κB and MAPK pathways. Data showed that the levels of p-Erk, p-p65, and IκBα were comparable; however, miR-145 mimics greatly repressed, by contrast, miR-145 inhibitor enhanced, the phosphorylation of MKK4, JNK, p38, c-Jun, and ATF2 under TNF-α stimulation (Figures 5a and b). Furthermore, miR-145 mimics additionally inhibited, whereas miR-145 inhibitor promoted, the nuclear import of p-c-Jun and p-ATF2 induced by TNF-α ( Figure 5c). It is worth noting that the facilitation of miR-145 inhibitor on TNF-α-induced upregulation of MMP-3, MMP-13, and Adamts-5 was partially blocked by JNK-and p38-specific inhibitors (Figure 5d (Figure 5f). In summary, these observations strongly suggest that miR-145 has a general inhibitory role in TNF-α-mediated JNK/c-Jun and p38/ATF2 activation and downstream gene induction.
MiR-145 attenuates cartilage matrix degradation in experimental OA likely through its suppression of MKK4-mediated TNF-α signaling. Considering that miR-145 has an inhibitory role in TNF-α-mediated signaling and induction of matrix-degrading enzymes in vitro, we further examined whether miR-145 affects catabolic genes expression and subsequent OA pathogenesis in vivo. For a more durable effect of miR-145 in vivo, we generated agomir-145 and antagomir-145 via several specific chemical modifications ( Figure 7a). Subsequently, agomir-145, antagomir-145, or an equivalent volume of vehicle was IA injected into the knee joint of rats with surgery-induced OA (Figure 7b).  (Figure 7f), indicating that MKK4-JNK/p38-c-Jun/ATF2-mediated suppression of matrixdegrading enzymes expression is mainly responsible for the observed inhibitory effect of miR-145 on cartilage destruction caused by OA (Figure 7g). Thus, our results support the utility of miR-145 as an effective therapeutic target for cartilage matrix degradation in OA.

Discussion
IL-1β and TNF-α are the major proinflammatory cytokines in OA. Several miRNAs involved in IL-1β-mediated signaling during OA have been reported. MiR-146a, an IL-1β-responsive miRNA, impairs TGF-β signaling pathway through the targeted inhibition of Smad4 in cartilage. 26 MiR-27b and miR-127 inhibit the IL-1β-induced upregulation of MMP-13 in human osteoarthritic chondrocytes. 24,30 However, no report on TNF-α-responsive miRNA in chondrocytes was found. In this study, we initially identified miR-145 as a TNF-αresponsive miRNA in chondrocytes, characterized by a rapid reduction upon TNF-α stimulation in vitro and a negative correlation with the secretion of TNF-α during experimental OA in vivo. Moreover, blocking the canonical NF-κB pathway or RNAi-mediated knockdown of tradd, traf2, or p65 partially rescued the reduction of miR-145 by TNF-α, suggesting that the inhibitory effect of TNF-α on miR-145 is mediated by p65. Previous studies have reported several miRNAs involved in OA pathogenesis. MiR-140, a cartilage-specific miRNA, regulates the expression of Adamts-5 in chondrocytes, 23 and miR-140-/-mice display an OA-like phenotype. 31 MiR-27a affects the expression of MMP-13 and IGFBP-5. 32 MiR-93 regulates collagen loss by targeting MMP-3, 33 and miR-125b regulates the expression of Adamts-4 in human chondrocytes. 34 Apparently, the OA-related miRNAs mentioned above merely target a single molecule among MMPs and ADAMTS. However, in the present study, we found that miR-145 overexpression could simultaneously suppress several matrix-degrading enzymes at transcriptional and post-transcriptional levels. In addition, neither overexpression nor inhibition of miR-145 affected the viability of chondrocytes. Thus, miR-145 may be a more effective target for OA treatment because of its broad inhibition of matrix-degrading enzymes production caused by TNF-α. Unexpectedly, miR-145 exerted no effect on anabolic factors in chondrocytes treated with TNFα. Consistently, our further studies confirm that the promoter regions of Mmp-3, Mmp-13 and Adamts-5 but not Sox-9, Col2a1, and Aggrecan, harbor the c-Jun or ATF2-binding motif; therefore, MMP-3, MMP-13, and Adamts-5 but not Sox-9, Col2a1, and Aggrecan can be directly modulated by the miR-145/MKK4-JNK/p38-c-Jun/ATF2 axis.
Our results demonstrated that miR-145 was sensitive to TNF-α, characterized by a rapid reduction upon TNF-α stimulation and a solid negative correlation with TNF-α secretion during the progression of human and experimental OA, implying that miR-145 is possibly involved in OA pathogenesis. Indeed, modulation of miR-145 efficiently affected ECM degradation during OA, as evidenced by that miR-145 gain-of-function was able to restrain TNF-α-induced matrix-degrading enzymes and cartilage destruction caused by surgery-induced OA. Moreover, we uncovered the molecular mechanism underlying the protective effect of miR-145 on OA cartilage in vitro and further confirmed it in vivo. Collectively, miR-145 critically controls OA development through strictly modulating the transcription of several key TNF-α-induced matrix-degrading enzymes. On the other hand, miR-145 had also been identified to be involved in the regulation of chondrogenic differentiation in a previous study. 35 Thus, we propose miR-145 as an important regulator of cartilage homeostasis, especially the catabolic signals, and its abnormal downregulation by TNF-α facilitates the progression of OA.
As is well known, the JNK and p38 MAPKs are activated by MKK4 and MKK7, and consequently capacitate the nuclear import of p-c-Jun and p-ATF2, eventually resulting in alteration of gene transcription. 36 However, no report was found on the involvement of MKK4 or MKK7 in the modulation of MMPs and ADAMTS during OA. In the present study, we report for the first time that MKK4 is a conserved target gene of miR-145. Binding of miR-145 to MKK4 3′-UTR severly impaired the translation of MKK4, resulting in suppression of TNF-αmediated signaling and induction of downstream matrixdegrading enzymes. Moreover, our results showed that MKK4 was significantly elevated in OA cartilage. Depletion of MKK4 or its upstream molecule HGK remarkably attenuated TNF-α-induced JNK/c-Jun and p38/ATF2 activation, and subsequently the production of matrix-degrading enzymes. In contrast, MKK4 overexpression accelerated TNF-αmediated signaling activation and induction of catabolic genes. In vivo, we observed that blocking JNK or p38 signaling, or depletion of MKK4 partially, attenuated cartilage destruction caused by surgery-induced OA.
TNF-α is an important mediator of matrix degradation during OA, exerting its functions through inducing and sustaining the inflammatory cascades. 4 Thus, negative regulation of TNF-α signaling is critical to shut off persistent inflammatory responses for the prevention of cartilage matrix degradation. So far, not much is known about how TNF-α-mediated signaling is negatively modulated. This study showed that miR-145 can be used as a potential novel target for therapy of OA because of its broadly inhibitory effects on TNF-αmediated signaling and expression of MMPs and ADAMTS in vitro and in vivo, and more importantly, compared with the recombinant proteins, miR-145 agonist via methylated modifications exerts more stable and durable effect. It is also worth mentioning that miR-145 may also serve as an efficient therapeutic target in inflammatory arthritis such as rheumatoid arthritis (RA) in consideration of the similar inflammatory conditions in patients with RA.
In conclusion, we elucidate a novel regulatory mechanism by discovering that miR-145 is a crucial negative regulator of TNF-α-mediated signaling activation and induction of cartilage matrix degradation mechanically through the MKK4-JNK/p38c-Jun/ATF2 axis during OA pathogenesis, and demonstrate the potential utility of miR-145 and MKK4 as therapy targets for OA. To our knowledge, miR-145 represents the first identified TNF-α-reduced miRNA, which in turn serves as a negative regulator of TNF-α-mediated signaling activation and transcription.
The role of microRNA-145 in cartilage degradation Guoli Hu et al

Materials and Methods
Human OA cartilage and experimental OA in rats. Human OA cartilage was sampled from OA patients (n = 10) who underwent total joint replacement. Undamaged areas in the same patient were sampled as the normal cartilage. Ethical approval was obtained from the Medical Ethics Committee of Xin Hua Hospital, Shanghai Jiao Tong University School of Medicine, and informed consent was obtained from all participants. All experiments involving rats were purchased from Shanghai SLAC Laboratory Animal Co. Ltd (Shanghai, China), and animal handling and experimental procedures were performed following the approval from the Institute of Health Sciences Institutional Animal Care and Use Committee. Experimental OA in 12-week-old SD rat (male, 200 g) was induced by medial collateral ligament transection and DMM as described previously. 37,38 SP600125 (JNK inhibitor, 10 μM) (S1460; Selleck, Houston, TX, USA), SB203508 (P38 inhibitor, 10 μM) (S1076; Selleck), CC-5013 (TNF-α secretion inhibitor, 10 μM) (S1029; Selleck), miR-145 agomir (50 μM), miR-145 antagomir (200 μM), and empty lentivirus (Len-C) or lentivirus (1 × 10 9 PFU, 20 μl) expressing sh-MKK4 (Len-sh-MKK4) or MKK4 (Len-MKK4) or the equivalent volume of vehicle (DMSO or PBS) were IA injected into the knee joints of recipient rats 1 week after the surgery (20 μl per joint per rat two times a week for 7 weeks) (n = 6-10 per group). Rats were killed 8 weeks after the surgery, and samples of the knee joints were collected.
Histology and immunostaining. Human OA cartilage was fixed in 4% paraformaldehyde and embedded in paraffin. Paraffin blocks were sectioned at a thickness of 5 μm. Sections were deparaffinized in xylene, hydrated with graded ethanol, and stained with HE. Cartilage destruction of rat knee joints was examined using safranin-O staining and scored by two observers blinded to group-identifying information using the OARSI grading system. 39 For immunohistochemistry, antigen retrieval was performed by incubating at 37°C with 0.05% trypsin (pH 7. Microarray. The miRNA expression profiles of the primary rat chondrocytes treated with or without TNF-α (10 ng/ml) for 24 h were determined by miRNA microarray analysis (LC Sciences, Houston, TX, USA). Data were analyzed by first subtracting the background and then normalizing the signals using a LOWESS filter (locally weighted regression). Normalized data were further analyzed by two-tailed Student's t-test. miRNAs with Po0.05 were considered differentially expressed (shown in Supplementary Table 2). In consideration of the modest elevated levels of upregulated miRNAs, we focused on the downregulated miRNAs. After assessing the cross-species conservation, five downregulated miRNAs (with maximal fold change) were selected for qRT-PCR validation. Finally, miR-145 was selected for further investigation because of its higher expression level in chondrocytes and cartilage compared with other miRNAs.
miRNA target prediction. The miRNA target interaction was predicted by miRanda. The TargetScan and miRDB were also applied to validate the interaction. Targets (MKK4, ROCK1, NEDD9, SOX9, and SMAD3) with high score (4100) were considered as candidate genes. The miRWalk database was further applied to exclude the genes that had been validated as targets of miR-145 in previous reports. The expression level of candidate genes in chondrocytes and cartilage were also evaluated by real-time PCR. Altogether, MKK4 was selected for further investigation.
Luciferase reporter assay. All plasmids for transfection were prepared using the Qiagen Plasmid Purification Kit (Qiagen, Hilden, Germany). The SW1353 cell line were co-transfected with miR-145 mimics or inhibitor (40 nM), luciferase constructs (described above) (200 ng), and pRL-TK (Promega) Renilla luciferase plasmid (50 ng). For promoter reporter assay, SW1353 cells were transfected with a mixture of 200 ng luciferase vector (containing p145-wt or p145-NF-κB-mut) and 50 ng pRL-TK Renilla luciferase plasmid. Luciferase assays were performed with the dual-luciferase reporter assay system (E1910; Promega) according to the manufacturer's instructions. Luminescent signals were quantified by a luminometer (Glomax; Promega), and each value from the Renilla luciferase construct was normalized by firefly luciferase. Real-time PCR. Total RNA from tissues or cells was extracted using Trizol reagent (Invitrogen). First-strand cDNA was synthesized from 1 mg of total RNA by incubating for 1 h at 42°C using the RevertAid Reverse Transcriptase (EP0442; Thermo Scientific, Waltham, MA, USA) following oligonucleotide (dT) priming in accordance with the manufacturer's instructions. qRT-PCR was performed by ViiATM 7 Real-Time PCR System (Life Technologies) using SYBR Premix Ex Taq (Takara, Dalian, China). Data were analyzed using the comparison Ct (2 − ΔΔCt ) method. GAPDH and small nuclear RNA U6 were used as internal controls for cDNA and miRNA, respectively. The data in Figure 4j was normalized by β-actin, due to the low uniformity of GAPDH among the different tissues. Each experiment was performed in triplicate. The primer sequences used in this study are summarized in Supplementary Table 4.
Chromatin immunoprecipitation. ChIP assay was performed using EZ ChIP Chromatin Immunoprecipitation Kit (Millipore) in accordance with the manufacturer's instructions. Immunoprecipitation was carried out overnight with p65 (8242 S; Cell Signaling Technology), c-Jun (9165; Cell Signaling Technology), and ATF2 (8638; Cell Signaling Technology) antibodies or normal rabbit IgG as a negative control. Protein A/G agarose was used to pull down the antigen-antibody compounds and then washed four times with washing buffers. The DNA-protein crosslinks were reversed with 5 M NaCl at 65°C for 6 h, and DNA from each sample was purified. PCR was performed with 2 μl of DNA samples with the primers shown in Supplementary  Table 4. The PCR products were analyzed by 2% agarose gel electrophoresis.
Enzyme-linked immunosorbent assay. The cytokine and the degradation products of collagen II (C2C) production from the samples of human OA cartilage and surgery-induced OA in rats were assessed with TNF-α (560479; BD Biosciences, San Jose, CA, USA) or C2C (YM8649; Yuan Mu Bioscience, Shanghai, China) ELISA Kits (enzyme-linked immunosorbent assay). A standard curve was generated using known concentrations of the respective purified recombinant TNF-α or C2C.
Statistical analysis. All statistical analyses were performed with SPSS 19.0 software (SPSS Inc., IBM Corporation, Armonk, NY, USA). Data are presented as mean ± S.D. Statistical differences between two groups were determined by two-tailed Student's t-test. Po0.05 was considered statistically significant.

Conflict of Interest
The authors declare no conflict of interest.