Zinc-induced protective effect for testicular ischemia-reperfusion injury by promoting antioxidation via microRNA-101-3p/Nrf2 pathway

Zinc prevented testicular ischemia-reperfusion (I/R) injury in rats

As shown in Figure 1A, the histopathological examination in I/R group showed congestion hemorrhage, germ cell loss, and disruption of the seminiferous epithelium, compared with Control group. In High-dose Zinc+I/R group, the pathological changes did not improve compared with I/R group. In contrast, the testicular tissues in Low-dose Zinc+I/R group showed an improved histological appearance compared to I/R group. As presented in Figure 1B, the MSTD in I/R group and High-dose Zinc+I/R group were significantly reduced compared to the values in Control group. In contrast, the MSTD in Low-dose Zinc+I/R group was significantly higher compared with I/R group (P<0.05). In addition, as shown in Figure 1C, compared with Control group, the number of testicular apoptotic cells in I/R group was significantly higher; Nevertheless, the percentage of apoptotic cells in Low-dose Zinc+I/R group was significantly lower than that of I/R group (P< 0.05), which provided evidence that Low-dose Zinc could protect against testicular I/R-induced apoptotic damage. Moreover, semi-quantitative analysis by both total TUNEL-positive cells/10³ germ cells and apoptotic index (AI) showed that Zinc could significantly decrease the incidence of testicular apoptosis and I/R-induced TUNEL-positive cells and AI (Figure 1D).

Figure 1. Zinc prevented testicular ischemia-reperfusion (I/R) injury in rats. (A) H&E staining of testicular tissues in Control, I/R injury, and I/R injury treated with Zinc rats (×400). Among them, ① Control group (n=8), ② I/R group (n=8), ③ Low-dose Zinc+I/R group (n=8), ④ High-dose Zinc group+I/R (n=8). (B) Mean seminiferous tubular diameter (MSTD) in each group. (C) TUNEL assay of testicular tissues in each group. Apoptotic cells exhibit a brown nuclear stain under microscope (×400). (D) TUNEL positive cells per 10³ germ cells and Apoptotic index of testes in each group. Data are shown as mean ± SD. *significant difference vs. Control group (P< 0.05); ^#significant difference vs. I/R group (P< 0.05).

Zinc alleviated testicular oxidative damage in rats

The intensity of the dihydroethidium (DHE) staining of testis in Control group was low, while it was markedly enhanced in I/R group. In contrast, the administration of Low-dose Zinc +I/R significantly reduced the intensity of the fluorescent signals, indicating that it could reduce testicular oxidative stress compared to I/R group (Figure 2A). As shown in Figure 2B–2G, the MDA content in I/R and High-dose Zinc+I/R groups were significantly higher compared to Control group (P<0.05); However, compared to I/R group, Low-dose Zinc+I/R group could significantly rescue MDA level (P<0.05). In addition, Low-dose Zinc+I/R group could increase the level of antioxidant enzymes SOD and CAT, and the content of antioxidants T-AOC and GSH, GSH/GSSG to reduce testicular oxidative stress injury compared to I/R group. Nevertheless, these parameters in I/R group and High-dose Zinc+I/R group were significantly lower compared to Control group, which indicated that Low-dose Zinc alleviated testicular oxidative damage in vivo.

Figure 2. Zinc alleviated I/R induced oxidative damage in testis tissues. (A) DHE staining of testicular tissues in Control, I/R injury, and I/R injury treated with Zinc group. ROS exhibited red fluorescence under fluorescent microscope. (B–G) Content of MDA, T-AOC, GSH, GSH/GSSG, GST and SOD of testicular tissues in each group. Data are expressed as mean±SD. *significant difference vs. Control group (P< 0.05); ^#significant difference vs. I/R group (P< 0.05).

Zinc pre-treatment increased the expression levels of Nrf2 and its downstream target genes in rats testicular I/R injury

The expression levels of Nrf2 and NF-κB in the testes as assessed from immunohistochemical analysis were as follows. Compared to Control group, higher level of NF-κB expression were detected in I/R group (P<0.05); Nevertheless, the expression level of NF-κB was decreased in Low-dose Zinc+I/R group, compared to I/R group (P<0.05). In addition, the increased Nrf2 expression was significantly observed in Low-dose Zinc+I/R group compared to I/R group (P<0.05) (Figure 3A–3B). As shown in Figure 3C, 3D, the protein levels of Nrf2, NF-κB, HO-1, NQO1 and GST were analyzed by Western blotting. The expression levels of Nrf2 and its downstream target genes in Low-dose Zinc+I/R group were significantly increased compared with I/R group (P<0.05). As expected, NF-kB expression level in Low-dose Zinc+I/R group significantly decreased compared to I/R group (P<0.05).

Figure 3. Zinc pre-treatment increased the expression levels of Nrf2 and its downstream target genes in testicular tissues. (A) Immunohistochemical Staining showed Nrf2 and NF-κB expression levels in Control, I/R injury and I/R injury treated with Zinc groups. (B) Statistical analyses of IHC results in Nrf2 and NF-κB expression levels in different groups. (C) Protein levels of Nrf2, NF-κB, HO-1, NQO-1 and GST in different groups. Histone H3 and β-actin were used as a protein control to normalize volume of protein expression. (D) Statistical analyses of Western Blotting results in Nrf2, NF-κB, HO-1, NQO-1 and GST protein levels in different groups. Data are expressed as mean± SD. *significant difference vs. Control group (P< 0.05); ^#significant difference vs. I/R group (P< 0.05).

Differentially expressed miRNAs in testicular tissues of rats induced by I/R injury based on microarrays

A total of 22 differentially expressed miRNAs with at least 2-fold changes and P-values less than 0.05 were identified in testicular tissues of rats caused by I/R injury using microRNA microarray analysis, compared with Control group. As shown in Figure 4A, the heat map indicated the results of a two-way hierarchical clustering of the samples and the 22 differentially expressed miRNAs, which displayed the relative expression levels identified by microarray. As shown in Figure 4B, the interaction networks between the Nrf2 and its upstream miRNAs were predicted by “miRDB”, “TargetScan” and “StarBase” databases. Then, 4 up-regulated and 1 down-regulated miRNAs were further validated by real-time PCR assay. As shown in Figure 4C, I/R injury significantly increased the expression levels of miR-101-3p, miR-144-3p, miR-140-5p, miR-17-5p, and markedly decreased the levels of miR-27b-3p in testicular tissues of rats compared with Control group. In addition, to investigate the the association between miR-101-3p and Nrf2 in TM3, miR-101-3p mimics and inhibitor was transfected into TM3. As shown in Figure 4D, miR-101-3p mimics significantly increased miR-101-3p in TM3, and miR-101-3p in TM3 was significantly decreased after administrated with miR-101-3p inhibitor. Then, Western Blotting showed that the protein level of Nrf2 decreased after administrated with miR-101-3p mimics; Nrf2 expression level increased after administrated with miR-101-3p inhibitor (Figure 4E).

Figure 4. Differentially expressed miRNAs in testicular tissues of rats induced by I/R injury based on microarrays. (A) Hierarchical clustering analysis of the differentially expressed miRNAs between Control and I/R injury testis samples in rats. (B) A schematic diagram used to search the target miRNAs of Nrf2 in three databases. (C) Validation of the five differently expressed miRNAs in rats caused by I/R injury based on real-time PCR assay. (D) MiRNA-101-3p expression level of TM3 after transfecting miR-101-3p mimics and inhibitor. (E) Nrf2 expression level of TM3 after transfecting miR-101-3p mimics and inhibitor. All data are expressed as the mean ± SD. *significant difference vs. Control group (P< 0.05).

Low-dose Zinc up-regulated Nrf2 signaling pathway via miR-101-3p in TM3 Hypoxia/Reoxygenation (H/R) model

As shown in Figure 5A, the cell viability was evaluated when combined with or without H/R. The results found that the cell viability was significantly decreased in H/R group. When TM3 were co-culture with Zinc, the cell viability was relieved, and Low-dose Zinc+H/R group was markedly better than High-dose Zinc+H/R group. Subsequently, the results of real-time PCR assay indicated that the expression levels of miR-101-3p was significantly down-regulated in TM3 H/R model (P<0.05); Meanwhile, compared to High-dose Zinc+H/R group, Low-dose Zinc pretreatment could markedly decrease the expression level of miR-101-3p (P<0.05) (Figure 5B). As shown in Figure 5C, the intracellular ROS levels in Zinc pretreatment group was significantly lower than H/R group based on immunofluorescence assays. In contrast, Low-dose Zinc pretreatment could markedly decrease the excessive generation of ROS compared to High-dose Zinc+H/R group. In addition, the intracellular Nrf2 in Zinc pretreatment group was also significantly higher than H/R group, and Low-dose Zinc pretreatment could markedly increase Nrf2 compared to High-dose Zinc+H/R group. Then, the protein levels of Nrf2, NF-κB and its downstream signaling molecules were assessed to observe Zinc-induced protective effect for TM3 H/R model. As shown in Figure 5D, the levels of Nrf2, HO-1, NQO1 and GST were markedly increased, and NF-κB were significantly decreased by Zinc pretreatment compared with H/R groups in vivo. Moreover, Low-dose Zinc pretreatment could markedly improve the above markers compared to the High-dose Zinc+H/R group, which indicated that Low-dose Zinc could activate Nrf2/HO-1 signaling pathways to protect TM3 from hypoxic damage.

Figure 5. Zinc up-regulated miR-101-3p via Nrf2/HO-1 signaling pathways in TM3 Hypoxia/Reoxygenation (H/R) model. (A) Cell viability of TM3 in Control, H/R injury, and H/R injury treated with Zinc groups. (B) MiRNA-101-3p expression level in H/R with or without Zinc pretreatment groups. (C) The intracellular ROS and Nrf2 levels in H/R with or without Zinc pretreatment groups. (D) Protein levels of Nrf2, NF-κB, HO-1, NQO-1 and GST of TM3 in different groups. Histone H3 and β-actin were used as a protein control to normalize volume of protein expression. (E) The construction diagram of the target genes (Nrf2) double-luciferase reporter genes. (F) The relative luciferase expression with Nrf2 3′-UTR after co-transfection with miR-101-3p mimics or NC mimics in TM3. Data are expressed as mean±SD. *significant difference vs. Control group (P< 0.05); ^#significant difference vs. H/R group (P< 0.05); **p < 0.01 compared with NC group; ns indicated no significance.

The RNA sequence alignment showed that the 3′-UTR of Nrf2 mRNA contained a complementary site for the seed region of miR-101-3p (Figure 5E). As presented in Figure 5F, the dual-luciferase reporter plasmid was obtained to perform the dual-luciferase reporter assay. In the groups of Nrf2-Wt, the luciferase activities were significantly repressed by miR-101-3p over-expression compared with mimics Control group. However, these effects were not observed with the mutated Nrf2 groups, suggesting that Nrf2 were the target genes of miR-101-3p.

Transfecting miR-101-3p mimics and inhibitor regulated H/R-induced oxidative damage in vitro

To investigate the role of miR-101-3p in TM3 H/R-induced oxidative damage, miR-101-3p mimics and inhibitor was transfected to TM3 in H/R model. As shown in Figure 6A, miR-101-3p mimics significantly increased miR-101-3p in TM3 H/R model, and miR-101-3p in TM3 H/R model was significantly decreased after administrated with miR-101-3p inhibitor. As presented in Figure 6B, the intracellular ROS in group transfected with miR-101-3p mimics was significantly higher than un-transfected group; However, the intracellular ROS in group transfected with miR-101-3p inhibitor was significantly lower than un-transfected group. In addition, the result of the intracellular Nrf2 was opposite after the same treatment. As shown in Figure 6C, the levels of Nrf2 and its downstream target genes HO-1, NQO1 and GST in TM3 H/R model were notably decreased and NF-κB was significantly increased after treating with miR-101-3p mimics compared with NC mimics group, and the results of protein levels were in contrast after treating with miR-101-3p inhibitor.

Figure 6. Transfecting miR-101-3p mimics and inhibitor regulated H/R-induced oxidative damage in vitro. (A) MiRNA-101-3p expression level of TM3 in H/R group after transfecting miR-101-3p mimics and inhibitor. (B) The intracellular ROS and Nrf2 of TM3 in H/R group after transfecting miR-101-3p mimics and inhibitor. (C) Protein levels of Nrf2, NF-κB, HO-1, NQO-1 and GST of TM3 after transfecting miR-101-3p mimics and inhibitor. Histone H3 and β-actin were used as a protein control to normalize volume of protein expression. Data are expressed as mean± SD. *significant difference vs. Control group (P< 0.05).

H/R regulated Nrf2 signaling pathway through miR-101-3p in TM3

As shown in Figure 7A, Nrf2 in TM3 cells was significantly decreased after administrated with si-Nrf2. Knockdown of miR-101-3p evaluated the cell viability in TM3 with H/R; Besides, si-Nrf2 further considerably reduced the effect of miR-101-3p inhibitor on the cell viability (Figure 7B). Knockdown of miR-101-3p via the transfection of miR-141-3p inhibitor reversed the H/R induced-promotion on NF-κB expression. Moreover, knockdown of miR-101-3p also reversed H/R induced-inhibition on the expression levels of Nrf2, HO-1, NQO1 and GST, while si-Nrf2 further considerably reduced the effect of miR-101-3p knockdown on Nrf2 and its downstream target genes (Figure 7C, 7D).

Figure 7. The interaction between miR-101-3p and Nrf2 in TM3. (A) Protein levels and statistical analyses of Western blotting results of Nrf2 in TM3 which administrated with si-Nrf2. (B) Cell viability of TM3 in different treated groups. (C) Protein levels of Nrf2, NF-κB, HO-1, NQO-1 and GST of TM3 in different groups. Histone H3 and β-actin were used as a protein control to normalize volume of protein expression. (D) Statistical analyses of Western blotting results in Nrf2, NF-κB, HO-1, NQO-1 and GST expression levels in different groups. Data are expressed as mean±SD. *significant difference vs. Control group (P< 0.05); ^#significant difference vs. H/R+NC inhibitor+si-Control group (P< 0.05); ^&significant difference vs. H/R+miR-101-3p inhibitor+si-Control group (P< 0.05).

Animals and experimental protocol

A total of 32 adolescent male Sprague-Dawley (SD) rats (180–220 g) at the age of averaging 6 weeks old were purchased from the Animal Experiment Center of Nanjing Medical University (Nanjing, Jiangsu, China). These rats were randomly divided into four groups: Control group (n=8), I/R group (n=8), High-dose Zinc+I/R group (60 mg/kg body weight) (n=8), and Low-dose Zinc+I/R group (15 mg/kg body weight) (n=8). The testes were examined via a scrotal incision and then replaced in the scrotum without torsion in Control group. In I/R group, the testes were rotated 720° in the clockwise direction for two hours, and then testicular reperfusion was performed by restoring the testes to their normal position. Subsequently, orchiectomy was performed four hours later. In High-dose Zinc+I/R group and Low-dose Zinc+I/R group, the same testicular torsion process was performed with Zinc (Santa Cruz, CA, USA) pre-treated for 14 consecutive days prior to testicular I/R injury surgery. The same volume of diluted solution was gavaged in Control and I/R groups. On the day of sacrifice, overnight fasted rats were anaesthetized with intraperitoneal injection of Pentobarbital (40 mg/kg BW).

All procedures conducted in experimental animals and the protocols were approved by the Committee on the Ethics of Animal Research in Animal Care Facility of Nanjing Medical University (Nanjing, Jiangsu, China). Unnecessary pain or stress was avoided and animal manipulation was performed with maximal care and hygiene. The study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health.

Surgical procedure

Surgical procedures were carried out by sterile techniques and performed under chloral hydrate anesthesia (300 mg/kg, intraperitoneally). Incising the ilioinguinal and exposing the testicle and spermatic cord. The left testis was rotated 720° in the clockwise direction for torsion in I/R group and Zinc-treated groups. To ensure this torsion position, we fixed the tunica albuginea to the scrotum with a 5/0 silk suture. Detorsion was operated by restoring the testes to their normal position after 2 hours of ischemia. Subsequently, orchiectomy was performed 4 hours later. The same surgical procedure was performed at the right testis on each rat. At the end, we collected testis tissues for Histopathological examination, Biochemical analysis, and Western blotting.

Histological examination

Testicular tissue samples were fixed in 10% formaldehyde for 24 hours, dehydrated, embedded in paraffin, sectioned at 5 μm, and stained with hematoxylin and eosin. Next, the sections were evaluated under a standard light microscopy (Olympus BX-51, Tokyo, Japan) for the change of testis structural by two blinded investigators. The mean seminiferous tubular diameter (MSTD) was measured in the same histological section on 5 different focuses with a microscope-adaptable micrometer.

TdT-mediated dUTP nick-end labeling (TUNEL) assay

TUNEL-staining assays were performed according to the manufacturer’s instructions (In Situ Apoptosis Detection Kit, Roche, Basel, Switzerland). The nucleus of any apoptotic cell would exhibit a brown stain under a standard fluorescence microscope, and this was manually analyzed in a blinded fashion. TUNEL-positive cells was presented per 10³ germ cells. On each slide, the apoptotic index was calculated by quantifying the number of TUNEL-positive versus total cell nuclei from five randomly chosen high-power fields (400×) containing 10³ cells.

Biochemical measurements

The level of Malondialdehyde (MDA), CAT, total antioxidant capacity (T-AOC), glutathione (GSH), SOD, and glutathione/glutathione disulfide ratio (GSH/GSSG) were determined by using an Assay Kit (Jiancheng Bioengineering Institute, Nanjing, China) [19, 23]. According to the method described by Benov et al. [24], we evaluated testicular intracellular ROS production by performing the intracellular superoxide assay under a standard fluorescence microscope (Eclipse Ti-SR, Nikon Co, Japan) and detected the density of the images by using a standard fluorescence spectrophotometer (arbitrary units per millimeter square field).

Immunohistochemical (IHC) staining

Immunohistochemistry was performed on paraffin testicle sections at 5 μm of thickness. For specific staining, we incubated slides with rabbit anti-mouse Nrf2 and NF-κB specific antibody (Cell Signaling Technology, USA) overnight at 4 °C. Washing and incubating the slides with a 1:100 dilution horseradish peroxidase (HRP)-conjugated anti-goat secondary antibody (Cell Signaling Technology, USA) for 1 hour the next day.

Cell culture and transfection

The Leydig cell (TM3) was purchased from the Shanghai Institutes for Biological Sciences (Shanghai, China). TM3 was maintained in DMEM supplemented with 10% FBS and antibiotics (100 IU/mL penicillin and 100 mg/mL streptomycin) in a humidified atmosphere of 5% CO₂ and 95% O₂ at 37°C. Prepare 50 μmol/L and 200 μmol/L Zinc-containing medium as Low-dose Zinc medium and High-dose Zinc medium, respectively. Transfection was performed to up-regulate or down-regulate the expression levels of miR-101-3p. Briefly, dissolving the miR-101-3p mimics, Negative Control mimics, miR-101-3p inhibitor or Negative Control inhibitor in Opti-MEM separately. The solutions were equilibrated for 5 mins at room temperature. Then, according to the manufacturer's protocol, combining each solution with Lipofectamine 3000 transfection reagent and mixing the solution gently allowed to form inhibitor liposomes for 20 mins. TM3 was transfected with the transfection mixture in serum-free cell medium. Place the mixed in a humidified atmosphere of 5% CO₂ and 95% O₂ at 37°C. Six hours later, replace the cell medium with fresh Medium. Finally, the expression levels of miR-101-3p and target genes was detected.

Cell hypoxic/reoxygenation (H/R) model

According to the manufacturer's protocol, cells were incubated under hypoxic conditions in a modular incubator chamber (Billumps-Rothenberg, Del Mar, CA). Place TM3 in the chamber flushed with a mixture of 94% N₂, 5% CO₂ and 1% O₂ at 37°C for 24 hours (oxygen-serum deprivation, ischemic cells) [45]. Next, return duplicate hypoxic cultures for reoxygenation in the normoxic incubator and serum-containing medium for 6 hours, and serve TM3 incubated under normoxic conditions as Controls.

Cell viability assay

According to the manufacturer's protocol, the cell viability with CCK-8 (Dojindo Laboratories, Kumamoto, Japan) was assessed. In brief, 10 μl CCK-8 solution was added to each plate and cells were incubated for 2 h at 37° after different treatments. Revealing the cell viability by the absorbance which was measured at 450 nm.

miRNA microarray

According to the manufacturer's instructions, total RNA samples from I/R group and Control group of rats (n=5) was isolated using TRIzol reagent (Invitrogen, USA) and purified using RNeasy mini kit (Qiagen, Germany). The hierarchical clustering was performed to show distinguishable miRNA expression profiling among samples.

Prediction of target miRNAs

The target miRNAs of Nrf2 validated above were screened by the “miRDB”, “TargetScan” and “StarBase” databases, and each target miRNA must be searched in at least two databases. Ultimately, the potential target gene was obtained.

Dual-luciferase reporter assay

Establishing the plasmids containing the wild-type miR-101-3p-Nrf2 (wt-Luc-Nrf2) response element and the corresponding mutant (mut-Luc-Nrf2), the blank plasmids were purchased from RiboBio. Co., Ltd. (Guangzhou, China). Co-transfecting Plasmid DNA (wt-Luc-Nrf2, mut-Luc-Nrf2) and miR-101-3p mimics or NC mimics into TM3. Finally, the luciferase activity was assessed with a Double-Luciferase Reporter Assay Kit which purchased from Promega Biotech Co., Ltd. (Beijing, China) and the Dual-Light Chemiluminescent Reporter Gene Assay System (Berthold, Germany) was used to firefly luciferase activity.

Immunofluorescence assay

Immunofluorescence staining in tissue sections or cells was performed using antibody in a humidified box at 4°C overnight, and followed by incubation with an Alexa fluorescein-labeled secondary antibody for 1 h at 37°C. The cell nuclei were stained with DAPI (5.0 μg/mL). Immunostained samples were imaged by fluorescence microscopy (Olympus, Japan) at 200× magnification.

The detection of differentially expressed miRNAs and mRNAs

According to the manufacturer's protocol, total RNA samples from TM3 and testicular tissues of rats were obtained in different treatments using TransZol. The SYBR Premix Ex TaqII of SYBR PrimeScript™ miRNA RT-PCR kit (Takara, Dalian, China) was used to reverse transcribed synthesis of cDNA to evaluate the expression level of miRNAs. Use SYBR PrimeScript™ miRNA RT-PCR kit to complete the following steps of qRT-PCR. Then, the MMLV Reverse Transcriptase First Strand cDNA Synthesis Kit (Invitrogen, Carlsbad, USA) was used to synthesize cDNA to assess the mRNA levels. qRT-PCR with SYBR Green Real-time PCR kit (Takara, Dalian, China) was performed on Applied Biosystems 7900HT Sequence Detection System (United States). We selected U6 and β-Actin as internal control for normalization of miRNAs and mRNAs, and analyzed the relative levels of gene expression by 2^−ΔΔCt method. In addition, each experiment was asked to repeat three times. The following primers were used for qRT-PCR: Nrf2, forward 5′-CTTTAGTCAGCGAC AGAAGGAC-3′; reverse 5′-AGGCATCTTGTTTGGG AATGTG-3′ NF-κB, forward 5′-CCAACAGATGGCC CATACCT-3′; reverse 5′-AACCTTTGCTGGTCCCA CAT-3′ Histone H3, forward 5′-CAAGCGCGTCAC TATCATGC-3′; reverse 5′-CCTTTGCTTGGACCGT CAGA-3′ HO-1, forward 5′-AGGTACACATCCAAGC CGAGA-3′; reverse 5′-CATCACCAGCTTAAAGCC TTCT-3′ NQO1, forward 5′-AGGATGGGAGGTACTC GAATC-3′; reverse 5′-TGCTAGAGATGACTCGGAA GG-3′ GST, forward 5′-TGCTGTCTTCTCTTCTTC CCAG-3′; reverse 5′-CCCAAAGGCTCCGTATCTG C-3′ β-Actin, forward 5′-CCTGGCACCCAGCACA AT-3′; reverse 5′-GCTGATCCACATCTGCTGGAA-3′.

Western blot analysis

According to the manufacturer's instructions, total protein was extracted from freshly obtained testicular tissues using a Nuclear Extract Kit (Active Motif, Tokyo, Japan). Primary antibodies against Nrf2, NF-κB, GST, HO-1 and NQO1 (Cell Signaling Technology, USA) were used in this study. We used Histone H3 and β-actin (Cell Signaling Technology, USA) as a protein control to normalize the volume of protein expression. The densitometric analysis of the protein bands was performed quantitatively with Image Lab Software (Bio-Rad, USA).

Statistical analysis

The experimental results were statistically analyzed by SPSS 22.0 (Armonk, New York, USA). The measurement data were expressed as the mean ± standard deviation (SD) for each group. It was performed with analysis of variance (ANOVA) followed by Dunnett's test when comparing multiple groups to evaluate the significance of differences between groups. Statistical significance was considered to be P < 0.05.