2020 Volume 43 Issue 5 Pages 864-872
Cardiac fibrosis is a major contributor for diabetic cardiomyopathy and Dendrobium officinale possessed therapeutic effects on hyperglycemia and diabetic cardiomyopathy. To further investigate the possible mechanisms of the Dendrobium officinale on diabetic myocardial fibrosis in mice. Water-soluble extracts of Dendrobium officinale (DOE) from dry stem was analyzed by HPLC and phenol-sulfuric acid method. Diabetic mice were induced by intraperitoneal injection of streptozotocin (STZ) (30 mg/kg) for 4 consecutive days after intragastric administration of a high-fat diet (HFD) for 2 weeks. The groups were as follows: control group, model group, DOE low, medium, high dose group (75, 150, 300 mg/kg) and Metformin positive group (125 mg/kg). The results showed that DOE dose-dependently lower serum insulin, total cholesterol (TC), triglyceride (TG), low-density lipoprotein cholesterol (LDL-C) and grew the high-density lipoprotein cholesterol (HDL-C) after 12 weeks of daily administration with DOE. Hematoxylin-eosin staining and Sirius red staining showed obvious amelioration of cardiac injury and fibrosis. In addition, the result of immunoblot indicated that DOE increased the expression of peroxisome proliferator activated receptor-α (PPAR-α), phosphorylation of insulin receptor substrate 1 (p-IRS1) and E-cadherin and repressed the expression of transforming growth factor β1 (TGF-β1), phosphorylation of c-Jun N-terminal kinase (p-JNK), Twist, Snail1 and Vimentin. The present findings suggested that DOE ameliorated HFD/STZ-induced diabetic cardiomyopathy (DCM). The possible mechanism mainly associated with DOE accelerating lipid transport, inhibiting insulin resistant and suppressing fibrosis induced by epithelial mesenchymal transition (EMT).
Diabetes mellitus is a serious threat to human health, and 451 million people were estimated to suffer from diabetes in 2017, more than 90% of whom are type 2 diabetes mellitus (T2DM), with the number increasing to 693 million in 2045.1) T2DM features chronic hyperglycemia with disturbances of fat metabolism which result from deficiency of insulin secretion and/or insulin resistance.2) Metabolic dysregulation caused by diabetes increases the risk of developing atherosclerosis, myocardial infarction, cardiomyopathy, and heart failure.3,4) Diabetes cardiomyopathy (DCM) is featured by impaired myocardial insulin signal, endoplasmic reticulum stress, mitochondrial dysfunction, sympathetic nervous system activation, excessive oxidative stress, aggravation of inflammation, abnormal coronary microcirculation and maladaptive immune response.5–8) These pathophysiological changes give rise to fibrosis and hypertrophy.9) Myocardial fibrosis is present recognized as the majority of DCM, giving rise to cardiac remodeling, cardiac dilatation and congestive heart failure.10)
Epithelial–mesenchymal transition (EMT) refers to the phenomenon that epithelial cells transform into interstitial cells, losing their epithelial cell characteristics under physiological and pathological conditions.11) There are three types of EMT, and type II occurs when specific cells are lost and replaced by fibrotic tissue in the course of a disease.12) Increasing evidence showed that EMT can lead to myocardial fibrosis due to non-synchronous heart failure through mechanical heterogeneity in the canine model.13) Fibroblasts in cardiac fibrosis are derived from endothelial cells.14) In patients with Crohn disease, an expression pattern of EMT was found in areas of fibrosis in the colon.15) Transforming growth factor β1 (TGF-β1) is an important profibrotic factor to induce EMT in fibrosis under physiological conditions. Novel therapeutic strategy targeting the interaction of TGF-β1 and EMT would effectively prevent cardiac fibrosis and slow the progression of DCM.
Dendrobium officinale Kimura et Migo (Dendrobium catenatum Lindley), a functional food and medicine herbal, has shown great pharmacological activity on diabetes and hypertension.16) Polysaccharides, the major constituent in dendrobium species,17) were recently reported to possess various potent pharmacological effects, including antioxidant, antiapoptotic, antitumor, and immunomodulation activity.18) The potential mechanism may be that pro-inflammatory cytokines, such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β and 619) and oxidative stress are inhibited.20) Our previous study suggested that oral administration of water-soluble exacts of Dendrobium officinale (DOE) could remarkably lowered blood glucose and prevent the development of DCM in streptozotocin (STZ) induced diabetic mice.21) However, the potential effects of Dendrobium officinale on cardiac fibrosis in diabetes still unclear. In the present study, we investigated whether DOE prevented insulin resistance and whether DOE reduced cardiac fibrosis through EMT in high-fat diet (HFD)/STZ-induced DCM mice.
The dried material of Dendrobium officinale Kimura et Migo (batch No: XZ20140301) were purchased from Xi’an Xiaocao Botanical Development Co., Ltd., Xi’an, China in January 2018 and authenticated by professor Zhubo Li (College of Pharmaceutical Sciences, Southwest University, Chongqing, China) in compliance with the identification standard of Pharmacopoeia of People’s Republic of China. The voucher specimens (No. 20140609) were submitted at the Herbarium of Materia Medica, Department of Traditional Chinese Medicine, College of Pharmaceutical Sciences, Southwest University, Chongqing, China. The dry stems were crushed into suitable power through 350-mesh. The powders were pre-extracted by petroleum ether and 80% ethanol with 60°C.21) The residues were extracted with double distilled water for 3 times, and thus the crude extracts were filtered, concentrated, dried by lyophilization. Forty seven gram powered extracts were collected from 200 g powders. The content of polysaccharides in DOE was determined by the phenol-sulfuric acid method. Glucose was used as the standard (Dglucose, Sigma, St. Louis, MO, U.S.A.).
Preparation of 1-Phenyl-3-methyl-5-pyrazolone (PMP) Derivatives of MonosaccharideThe monosaccharide composition in DOE polysaccharide was analyzed by HPLC. The amount of mannose and glucose were determined by PMP pre-column derivatization method on the basis of the Pharmacopoeia of the People’s Republic of China (2015 Edition). As described by Xiang,18) PMP (Sigma) derivatives of mannose, glucose, galactose, galacturonic acid and arabinose (National Institute for the Control of Pharmaceutical and Biological Products, Beijing, China) were prepared before HPLC analysis.
HPLC AnalysisThe RPHPLC (LC 20A, SHIMADZU, Japan) system equipped with Dikma Diamonsil C18 column (150 × 4.6 mm; 5 mm; Dikma Technologies, China) and SPD-20 A detector was performed to separate the PMP-derivatives of monosaccharide. The flow phase is composed of 80% ammonium acetate (A, 0.02 M) and 20% acetonitrile (B). The wavelength of the detector was 250 nm and the column temperature was 30°C. Internal standard method was performed to the quantitative analysis. The amount of mannose and glucose was expressed as percentage of the extracts of Dendrobium officinale.
Animals and DOE TreatmentKunming male mice, 8–10 weeks of age, weighting 20 ± 2 g, were purchased from Chongqing Tengxin Biotechnology Co., Ltd. [SCXK (Yu) 2017-0002]. The mice were treated with 22°C with a 12 h light/dark cycle and free access to food and tap water. All the animal were administrated according to the National Institutes of Health (NIH) guidelines and approved by the Ethical Committee for Animal of Southwest University.
The diabetic mice model was induced by intraperitoneal injection of streptozotocin (Sigma) at the dose of 30 mg/kg body weight (freshly dissolved in 0.1 M sodium citrate buffer pH 4.5) for 4 consecutive days after intragastric administration of HFD (19.67 kJ/g, 45% of energy from fat, 35.2% of energy from carbohydrate, 19.8% of energy from protein, purchased from Botai Hongda Biotechnology Ltd., Beijing) for 2 weeks.22) Blood glucose levels were measured using glucometer (Sinocare Inc Co., Ltd., Changsha, China) by tail vein puncture blood sampling, mice with blood glucose values >11.1 mmol/L were used for the study. The age-matched mice were randomly divided into the control group (n = 10) and diabetic mice group (n = 50). The control mice were received normal diet and multiple injections of the same volume of sodium citrate buffer. The diabetic mice were randomly divided into five groups: model group (normal saline), DOE low, medium, high dose group (75, 150, 300 mg/kg) and Metformin positive control group (125 mg/kg). These mice were received normal saline or DOE or Metformin gavaged once a day for 12 weeks, with the body weight being recorded. Then, the mice were euthanized. blood samples and heart and liver tissue were collected.
Measurement of Fasting Blood Glucose, Serum Insulin, Triglyceride (TG), Total Cholesterol (TC), High-Density Lipoprotein Cholesterol (HDL-C) and Low-Density Lipoprotein Cholesterol (LDL-C)Blood samples were obtained from the tails of the mice. The fasting blood glucose levels were measured with a glucometer once a week. The mice fasted for 12 h at a time. After treatment with DOE for 12 weeks, mice were sacrificed and blood samples were collected and centrifuged at 3000 rpm for 10 min at 4°C. And then collect the upper serum. The serum insulin, TG, TC, HDL-C and LDL-C were measured in accordance with the instructions of the commercial kit (Jiancheng Institute of Biotechnology, Nanjing, China).
Estimation of Insulin ResistanceHomeostatic index of insulin resistance (HOMA-IR) is a useful indicator for diagnosing insulin resistance.23) and the HOMA-IR was calculated by the formula fasting glucose (mmol/L) × fasting insulin (μU/mL)/22.5.24)
Hematoxylin–Eosin (H&E) StainingHeart tissues were obtained and fixed in 4% paraformaldehyde solution for 48 h. Heart tissues were then embedded in paraffin. The 5 µm-thick serial sections cut from paraffin blocks were stained with H&E trichrome solutions. Images of these stained sections were obtained using a light microscope.
Sirius Red StainingHeart tissues were obtained and fixed in a 4% paraformaldehyde solution for 48 h. Heart tissues were then embedded in paraffin. Four micrometers cross-sections cut from paraffin blocks were stained with 0.1% picrosirius red, and photographed by a light microscopy. The red areas of collagen deposition were counted by Image Pro-plus (version 6.0).
Western Blotting AnalysisTotal proteins were extracted from heart tissues with RIPA pyrolysis buffer (phenylmethylsulfonyl fluoride (PMSF), phosphatase inhibitors) and loading buffer. Equal masses of total protein were loaded on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and then transferred onto a 0.45 µm polyvinylidene difluoride (PVDF) membrane. Following blocking in 5% milk for 2 h, the membranes were incubated with primary antibodies against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1 : 1000, Wanlei, China), peroxisome proliferator activated receptor-α (PPAR-α) (1 : 1000, Wanlei, China), phosphorylated c-Jun N-terminal kinase (p-JNK) (1 : 1500, Wanlei, China), JNK (1 : 1000, Wanlei, China), p-insulin receptor substrate 1 (IRS1) (1 : 800, Amyjet Scientific, China), IRS1 (1 : 500, Wanlei, China), Twist (1 : 500, Wanlei, China), Snail 1 (1 : 1000, Wanlei, China), E-cadherin (1 : 1000, Wanlei, China), Vimentin (1 : 1000, Wanlei, China) overnight at 4°C. Then the membranes were further reacted with appropriate HRP goat anti-rabbit immunoglobulin G (IgG) (1 : 5000, Antgene Biotech) for 1 h. The chemiluminescence signals were recognized by ECL reagents (Advansta, CA, U.S.A.). Blots were visualized with Image software.
Statistical AnalysisSPSS 16.0 software (SPSS, Inc., Chicago, IL, U.S.A.) was performed to process all data presented as mean ± standard deviation (S.D.). Statistical analyses of the data were performed by one-way ANOVA using post hoc multiple comparisons; p < 0.05 was considered a significant difference.
High sensitivity and high consistency calibration curves were performed for glucose. Polysaccharide is the main component of water extract and the concentration of polysaccharide determined by phenol-sulfuric acid method was 44.83%. As shown in the result of RP-HPLC (Fig. 1), the PMP-labeled monosaccharide in DOE were separated by HPLC. The result manifested that DOE contains two major monosaccharides, including mannose and glucose. And the content of mannose and glucose is 28.97 and 4.6%, respectively.
(A) A mixture of reference substances: PMP (peak 1), mannose (peak 2), glucosamine hydrochloride (peak 3), galacturonic acid (peak 4), glucose (peak5), galactose (peak 6), arabinose (peak 7). (B) The species of monosaccharide in DOE: mannose (peak 2), glucose (peak 5).
After the model establishment, body weight of all diabetes mice was similar and was significantly lower than that in normal mice (Fig. 2A). After 12 weeks of treatment with DOE, a significant increase in body weight was observed. The body weight in high dose DOE-treatment group increased from 36.12 ± 1.90 to 39.98 ± 3.08 g. In addition, the heart weight and heart-to-body weight ratio (HW/BW) in the diabetic groups were significantly higher than that in the control group. Treatment with DOE reversed their upward trend, and the HW/BW of high dose group was remarkably lower than the model group (Figs. 2B, 2C). The data manifested that DOE may protect heart against cardiac hypertrophy.
(A) Body weight. (B) Heart weight. (C) Heart/body weight ratio. Date are expressed as mean ± S.D. (each n = 6). * p < 0.05 vs. control group. # p < 0.05 vs. STZ group.
After the model establishment, the fasting blood glucose of all diabetes mice were similar and was significantly higher than that in normal mice (Fig. 3A). But the fasting blood glucose levels of all diabetes model kept falling after 12 weeks of gastrointestinal treatment with DOE. The results of the last fasting blood glucose test showed that treatment with DOE reversed their upward trend, and the fasting blood glucose levels of middle and high dose group was remarkably lower than the model group (Fig. 3B). Insulin resistance and abnormal secretion are central to the development of type 2 diabetes. As illustrated in Fig. 3C, the serum fasting insulin of all diabetes model was higher that in normal mouse, and that in high dose group significantly lower than that in model group. The HOMA-IR index was 4-fold higher in the diabetes group than in the control group (Fig. 3D), nevertheless that in the low DOE, middle DOE and high DOE groups was decreased 3-, 2.8- and 2-fold, respectively. These data manifested that DOE distinctly ameliorated fasting blood glucose and relieved insulin resistance.
(A) Changes in serum fasting glucose during 12 weeks of continuous administration. (B) Fasting blood glucose. (C) Serum fasting insulin. (D) HOMA-IR. Date are expressed as mean ± S.D. (each n = 6). * p < 0.05 vs. control group. # p < 0.05 vs. STZ group.
To investigate whether DOE treatment could increase fatty acid metabolism in HFD/STZ-induced DCM mice. We measured TC, TG, HDL-C and LDL-C levels in the serum. As showed in Fig. 4, TC, TG, HDL-C and LDL-C levels in the diabetes model group were 4.61 ± 0.30, 3.81 ± 0.07, 1.65 ± 0.21 and 2.79 ± 0.18 (mmol/L), respectively, while they in the control group were 3.43 ± 0.29, 1.85 ± 0.12, 2.97 ± 0.19 (mmol/L) and 1.45 ± 0.07 (mmol/L), respectively. Obviously, TC TG and LDL-C increased and HDL-C decreased in diabetic mice compared to that in control group. After treatment with DOE for 12 weeks, the TC level significantly decreased to 4.49 ± 0.18, 4.30 ± 0.10 and 4.00 ± 0.33 (mmol/L) in low, middle and high groups, respectively. The TG level decreased to 3.35 ± 0.48, 2.94 ± 0.25 and 2.44 ± 0.15 (mmol/L) in low, middle and high groups, respectively. The HDL-C and levels increased to 1.85 ± 0.12, 2.10 ± 0.10, 2.69 ± 0.32 (mmol/L), respectively. Moreover, the LDL-C decreased to 2.51 ± 0.09, 2.00 ± 0.10, 1.78 ± 0.19 (mmol/L), respectively. In DOE groups, TC, TG and LDL-C levels were significantly reduced and HDL-C was significantly increased compared with dyslipidemic-diabetic mice. In addition, Western blotting results revealed that DOE dose-dependently up-regulated the level of PPAR-α and p-IRS1 and down-regulated the level of p-JNK in heart and liver tissue (Fig. 5). The data illustrated that DOE ameliorated fatty acid metabolism via PPAR-α and JNK in HFD/STZ-induced DCM mice.
(A) TC. (B) TG. (C) HDL-C. (D) LDL-C. Date are expressed as mean ± S.D. (each n = 6). * p < 0.05 vs. control group. #p < 0.05 vs. STZ group.
The PPAR-α, p-JNK, JNK, p-IRS1 and IRS1 in cardiac and liver tissue were assessed by Western blot analysis. Date are expressed as mean ± S.D. (each n = 6). *p < 0.05 vs. control group. # p < 0.05 vs. STZ group.
H&E staining was performed to evaluate the myocardial injury in HFD/STZ-induced DCM mice. As shown in the control group, the left ventricles structural was clear and well organized, and morphological changes were not detected (Fig. 6). By contrast, perinuclear vacuolization, necrosis and inflammatory infiltration were obvious in STZ group. DOE treatment normalized changes in heart tissue. In the DOE group, there were less abnormal myocardial structures, such as necrosis, cavitation and loss of muscle fibers under the microscope. The date illustrated that DOE protects against cardiac injury in HFD/STZ-induced DCM mice.
The arrows indicate cytoplasmic vacuolization, cardiomyocyte necrosis and inflammatory infiltration. Paraffin-embedded cardiac tissue stained with H&E were observed under light microscope (200×, bar = 100 µm).
The development of interstitial fibrosis is also a structural hallmark of diabetic cardiomyopathy. To measure the cardiac fibrosis, Sirius-red staining was used to determine the collagen represented by the red areas in the myocardial tissues (Fig. 7A). Obviously, the red areas of the control groups were the least. As shown in Fig. 7B, the deposition of collagen was 4.5-fold higher in the diabetes group than in the control group, whereas that in the low DOE, middle DOE and high DOE groups was decreased to 3-, 2- and 1.5-fold, respectively. The data demonstrated that DOE obviously decreased the deposition of collagen, downgraded of fibronectin and ameliorated fibrosis in HFD/STZ-induced DCM mice.
(A) Paraffin-embedded cardiac tissue was stained with Sirius red staining. As indicated by the arrows, the red area represents the fibrosis in cardiac tissue (200×, bar = 100 µm). (B) The red area was measured by semi-quantitative analysis. Date are expressed as mean ± S.D. (each n = 4). * p < 0.05 vs. control group. # p < 0.05 vs. STZ group.
Western blotting analysis was performed to determine the progress of EMT. The expression of Vimentin was increased and the expression of E-cadherin decreased in heart tissue of diabetes groups. Of note, the downregulation of epithelial cell junction proteins E-cadherin and the activation of mesenchymal adhesion genes vimentin were characteristic of EMT. DOE treatment increased the expression of E-cadherin and decreased the expression of Vimentin. Meanwhile, the increased expression of twist and snail 1, transcription regulators of EMT, also confirmed the enrichment of EMT characteristics in model group (Fig. 8A). In addition, TGF-β1, a potent inducer of EMT was shown obvious dose-dependent down-regulation in heart tissue after treatment with DOE (Fig. 8B). Those result demonstrated that DOE attenuated EMT in HFD/STZ-induced DCM mice.
TGF-β1, Twist, snail1, E-cadherin and Vimentin in cardiac tissue were assessed by Western blot analysis. Date are expressed as mean ± S.D. (each n = 6). * p < 0.05 vs. control group. # p < 0.05 vs. STZ group.
In the present study, diabetic mice exhibited symptoms of excessive intake, excessive excretion and emaciation. Treatment with DOE for 12 weeks relieved these symptoms and decreased fasting blood glucose in DCM mice. The biochemical indexes indicated that DOE reduced the level of TC, TG, LDL-C and serum fasting insulin and increased level of HDL-C in DCM mice. The results of HOMA-IR calculation showed that DOE relieved insulin resistance. Hematoxylin-eosin staining and Sirius red staining showed that DOE remarkably decreased cardiac injury and fibrosis. Moreover, Western blotting assay revealed that DOE ameliorated lipid transport and suppressed EMT in HFD/STZ-induced DCM mice.
Insulin-producing pancreatic endocrine cells are selectively destroyed by STZ. Long periods of hyperglycemia lead to changes in the pathology and function of various organs,25) such as heart and liver. Elevated blood glucose level, increased drinking water, more food consumption, and blood urea nitrogen production, as well as reduced body weight were shown in diabetes mellitus mice.26) Type 2 diabetes mellitus is characterized by insulin secretory dysfunction and insulin resistance. The clinical prevention and treatment of type 2 diabetes mellitus mainly begins with the reduction of insulin resistance and insulin secretory dysfunction.27) HOMA-IR is not only a useful indicator for diagnosis of insulin resistance, but also a follow-up indicator for the treatment of type 2 diabetes.28) In the present experiment, DOE significantly increased body weights and decreased HOMA-IR in HFD/STZ induced diabetic mice. In addition, peroxisome proliferator-activated receptors is an essential roles in glucose and lipid metabolic processes.29) Easily binding polyunsaturated fatty acids, PPAR-α accelerates β-oxidation of adipocytes in islet β cells and the clearance of fat in insulin-sensitive organs, resulting in increased insulin secretion under glucose stimulation.30,31) It reported that the protection of mice with macrophage-specific JNK deficiency against insulin resistance was associated with reduced tissue infiltration by macrophages.32) The JNK signaling pathway regulates the PPARα-FGF21 hormone axis.33) Sustained JNK activity is known to contribute to endoplasmic reticulum stress.34) The inhibitory serine phosphorylation of IRS-1 by JNK is known to underline inflammatory-as well as free fatty acid-induced insulin resistance.35) The study indicated that DOE significantly reduces TC, TG, HDL-C and LDL-C in DOE treatment group. Meanwhile, the increased expression of PPAR-α and p-IRS1 and the decreased expression of p-JNK were also found in the present study. In conclusion, the hypoglycemic and hypolipidaemic effects of DOE may prevent the deleterious effects of hyperglycemia and hyperlipidemia on the development of diabetes and diabetic cardiomyopathy. The possible mechanism may be associated with the activation of PPAR-α/JNK pathway.
The development of fibrosis is one of the structural hallmark and the major causes of diabetes cardiomyopathy.36) Sirius red staining showed that DOE can significantly inhibit fibrosis. The underlying mechanism might be that pro-fibrotic factor TGF-β37) was inhibited. In the present study we have confirmed that DOE indeed reduced the expression of TGF-β1 protein. Under long-term hyperglycemia conditions, TGF-β1 is activated and the activated TGF-β1 is directly associated with TGF-β receptor II, which raises TGF-β receptor I and lead to phosphorylation of Smad2 and Smad3.38) The activated phosphorylation of Smad2 and Smad3 could promote the deposition of extracellular matrix accumulation proteins, including collagen, elastin, laminin and fibronectin.39) In addition, TGF-β is a potent inducer of EMT, which has been reported by numerous studies.40) Epithelial cells EMT-differentiated transform into mesenchymal phenotypes, producing fibroblasts and myofibroblasts and EMT is widely considered as playing an important role in fibrosis.41) Partial EMT process after epithelial cell injury leads to prolonged cell proliferation, cell cycle stagnation, and secretion of fibrogenic factors, thereby promoting fibrosis and parenchymal injury.42) The Snail family and Twist family were two major groups of EMT-activating transcription factors, which are used to demonstrate the functional significance of EMT.43) During the epithelial mesenchymal process, epithelial polarity and cell connections are absent, while E-cadherin maintains tight junction in cells.8) Similarly, Vimentin is activated and intensifies epithelial mesenchymal fibrosis.44) The current study indicated that the expression of E-cadherin was significantly decreased, while Vimentin, Twist and Snail1 levels were remarkably increased in the model mice. However, DOE up-regulated the expression of E-cadherin and down-regulated the expression of Vimentin, Twist and Snail1 in cardiac tissue. The results demonstrated that DOE rescued EMT and reduce the progression of myocardial fibrosis.
In conclusion, the present study demonstrated that oral administration of DOE effectively ameliorated HFD/STZ induced DCM by accelerating lipid transport, inhibiting insulin resistant. Furthermore, DOE suppressed myocardial fibrosis through inhibiting epithelial mesenchymal transition in DCM mice. Herein, the results indicated that DOE possessed heart-protective effects against DCM and can serve as a potential drug for treating DCM.
The present study was supported by the Chongqing Research Program of Basic Research and Frontier Technology, China (Grant No. cstc2016jcyjA0296), the Fundamental Research Funds for the Central University (XDJK2019B058) and the Postdoctoral Science Foundation of Chongqing (No. Xm2017083).
Xiaoyan Zhao and Zhubo Li designed the project; Dongning Li performed the experiments; Dongning Li and Jie Zhang wrote the manuscripts; Jie Zeng analyzed and interpreted data; all authors reviewed the manuscript.
The authors declare no conflict of interest.
