Identification of Neuroendocrine Stress Response-Related Circulating MicroRNAs as Biomarkers for Type 2 Diabetes Mellitus and Insulin Resistance

Background Chronic stress plays an important role in the development of type 2 diabetes mellitus (T2DM) and insulin resistance (IR). MicroRNAs (miRNAs) play key roles in mediating stress responses by regulating the expression of target genes. This study systematically screened and identified the neuroendocrine stress response-related circulating miRNAs which are associated with T2DM and IR. Methods Based on the differential plasma expression profiles between individuals with and without T2DM, stress-related miRNAs were selected from those differently expressed miRNAs whose targets are involved in known neuroendocrine pathway of stress response. Candidate miRNAs were further validated by quantitative real-time polymerase chain reaction in a large sample, including 112 T2DM patients, 72 individuals with impaired fasting glucose (IFG), and 94 healthy controls. The association between miRNA expression and potential risk of T2DM and IFG was assessed by multivariate logistic regression models. The miRNA predictors of IR were identified by stepwise multiple regression analysis. The diagnostic performance for T2DM was evaluated by area under the curve (AUC) of receiver operating characteristic (ROC). Results let-7b, let-7i, miR-142, miR-144, miR-155, and miR-29a were selected as candidate miRNAs for validation. Increased expression of let-7b, miR-144, and miR-29a and decreased expression of miR-142 were significant independent predictors of T2DM, IFG, and IR (P < 0.0125). These miRNAs significantly correlated with stress hormone levels (P < 0.0125). A three-miRNA panel, including let-7b, miR-142, and miR-144 had a high accuracy for diagnosing T2DM (AUC = 0.871, 95% CI: 0.822–0.919). Conclusion let-7b, miR-142, miR-144, and miR-29a in plasma may be important markers of neuroendocrine stress response and may play a role in the pathogenesis of T2DM and IR.

Methods: Based on the differential plasma expression profiles between individuals with and without T2DM, stress-related miRNAs were selected from those differently expressed miRNAs whose targets are involved in known neuroendocrine pathway of stress response. Candidate miRNAs were further validated by quantitative realtime polymerase chain reaction in a large sample, including 112 T2DM patients, 72 individuals with impaired fasting glucose (IFG), and 94 healthy controls. The association between miRNA expression and potential risk of T2DM and IFG was assessed by multivariate logistic regression models. The miRNA predictors of IR were identified by stepwise multiple regression analysis. The diagnostic performance for T2DM was evaluated by area under the curve (AUC) of receiver operating characteristic (ROC).
conclusion: let-7b, miR-142, miR-144, and miR-29a in plasma may be important markers of neuroendocrine stress response and may play a role in the pathogenesis of T2DM and IR.
Keywords: neuroendocrine, stress, microrna, type 2 diabetes mellitus, insulin resistance, biomarker Stress-Related miRNAs as Biomarkers for T2DM/IR Frontiers in Endocrinology | www.frontiersin.org March 2018 | Volume 9 | Article 132 inTrODUcTiOn Type 2 diabetes mellitus (T2DM) and its complications have been identified as a major public health problem with a great impact on global morbidity and mortality, and heavy economic burden in both developed and developing countries (1). In 2015, 415 mil lion people had diabetes and the number is predicted to increase to 642 million by 2040 (2). With increasing urbanization and strong economic growth in modern society, psychosocial stress has now emerged as an important contributor to the development of metabolic disorders. Chronic stress has been shown to cause impaired glucose homeostasis in earlier experimental and human researches (3). Recent epidemiological studies have confirmed the association between psychological stress and increasing risk of insulin resistance (IR), central obesity, and T2DM (4)(5)(6). Persistent psychological stress can affect health through neuro endocrine pathways including hypothalamuspituitaryadrenal (HPA) axis and sympathetic nervous system (SNS). The catecho lamines epinephrine (E) and norepinephrine (NE), produced by the SNS, and cortisol, produced by the HPA axis, are the major stress hormones and important mediators between psycho logical stress and impaired glucose and lipid metabolism. Several researches have shown that increased activation of HPA axis and SNS, and consequent elevation of hormones might be responsible for the etiology and development of T2DM (7,8).
MicroRNAs (miRNAs) are small noncoding regulatory RNAs which have proven powerful tools in finding disease associated pathways. They primarily regulate the expression of target genes at the posttranscriptional level by inhibiting translation or caus ing degradation of the target mRNA (9). Emerging data suggest that stress conditions can alter the biogenesis of miRNAs, which in turn determines cellular processes in response to stresses (10). Acute and chronic immobilization stress differentially altered the expression of miR183 and miR134 in stressresponsive regions of the rat brain (11). Both miR134 and miR183 were upregulated under acute stress in the central amygdala, while miR134 showed decreasing expression with chronic stress in both the central amygdala and the hippocampus CA1 region. Increased serum miR29c expression was identified after an acute social stress task in healthy participants (12). Levels of seven miRNAs (miR16, 20b, 26b, 29a, 126, 1445p, and 1443p) in peripheral blood were significantly elevated in association with significant down regulation of their target mRNAs with chronic academic stress (13). Previous investigation in healthy occupational population showed that psychological stress increases plasma cortisol, which subsequently impacts expression of glucocorticoid receptor (GR) in peripheral blood mononuclear cells (14). Based on the above evidences, we assume that miRNAs may involve neuroendocrine pathway of stress responses and contribute to the susceptibility of T2DM.
In addition to their typical intracellular function, miRNAs also participate in celltocell communication by actively secreting into the extracellular media, including serum, plasma, and small lipid vesicles, such as microvescicles and/or exosomes (15)(16)(17). Circulating miRNAs found in blood change with the physiologi cal condition of the organism and may help to predict, diagnose, and followup metabolic disease (18). Therefore, circulating miRNAs may serve as valuable biomarkers and providing thera peutic benefits for T2DM.
In this study, it is the first time that the association between neuroendocrine stress responserelated circulating miRNAs and T2DM was explored in an occupational population. The reason for our choice of such a population was that work environment issues are considered to be one of the most important psychoso cial stress in contemporary societies, especially in urban China (14,19). Based on an integrative analysis of differential miRNA expression profiles between T2DM patients and matched control subjects and target prediction, six candidate miRNAs that may involve in regulation of neuroendocrine stress response were selected. To reveal the role of these miRNAs in the development of T2DM, we also examined their expression changes in predia betes and their association with insulin resistance.

MaTerials anD MeThODs study Design
This study was performed based on a threestage design. First, we compared the plasma miRNA profiles of individuals from an occupational population with and without T2DM. Neuro endocrine stress responserelated miRNAs were then selected from those differently expressed miRNAs whose targets are involved in known biological pathways of stress response, includ ing HPA axis and SNS. A large sample was eventually used to validate the expression of candidate miRNAs by quantitative realtime polymerase chain reaction (qRTPCR).

subjects
The subjects in this study were Han Chinese individuals from a functional community cohort which was established in 2010 (20). This cohort was composed of nearly 9,000 employees from 64 companies (including whitecollar and bluecollar workers from governments, schools, hospitals, factories, business, and service institutions) in Xuanwu district, representing most of the occu pational population in urban Beijing. All the individuals in the cohort took annual physical examination at the health medical examination center of Beijing Xuanwu Hospital, Capital Medical University. The inclusion and exclusion criteria of the sample were described previously (20).
For the first stage, 10 T2DM patients and 10 individual matched controls (gender and age difference <12 months) were selected from the cohort in 2015 to compare the expression profile of miRNAs. At the validation stage, all the 112 newly diagnosed T2DM cases aged from 30-65 years old during 2017 were recruited. 94 health individuals with normal fasting plasma glucose (FPG < 6.1 mmol/l) were selected as frequency matched controls according to gender and age (±3 years). Another 72 indi viduals with impaired fasting glucose (IFG) were also recruited as prediabetes cases. The 1999 World Health Organization diagnostic criteria were used to diagnose T2DM (fasting glu cose ≥ 7.0 mmol/l and/or 2 h glucose ≥ 11.1 mmol/l) and IFG (fasting glucose ≥ 6.1 and <7.0 mmol/l).
Each participant signed an informed consent form and this study was approved by the Ethical Committee of Capital Medical University.

Data collection and anthropometric Measurement
Information about demographic data, medical history, current medication, and environmental exposure history were collected by a structured questionnaire (21). Anthropometric parameters, including weight, height, waist circumference (WC), and blood pressure were obtained using standard measurement described previously (6). Body mass index (BMI) was calculated as weight (kg) divided by height squared (m 2 ).

Plasma collection
Following an overnight fast, peripheral blood (5 ml) was col lected using EDTAcontaining tubes between 7:30 and 8:30 a.m. Samples were immediately centrifuged at 2,000 r/s for 10 min at 4°C, generating the onestep centrifugation plasma samples. Then, the upper plasma phase was transferred to 1.5 ml tubes and centrifuged at 13,000 r/s for another 10 min at 4°C to remove additional cellular debris and minimize contamination of cell free nucleic acids (gDNA and RNA) from damaged blood cells. The twostep centrifugated plasma samples were then stored at −80°C until further analysis.
Plasma adrenaline (E), norepinephrine (NE), cortisol, and insulin were measured by commercial radioimmunoassays using with a γ counter (XH6020, North Institute of BioTech, China). Plasma corticotropinreleasing hormone (CRH), interleukin6 (IL6) concentrations were determined by enzyme linked immuno sorbent assay using microplate reader (STAT FAX 2100, Awareness, USA). The coefficient of variation of these assays is <6.0% for the intraassay and <10.0% for the interassay, respectively. The degree of insulin resistance was evaluated using the homeostasis model assessment of insulin resistance (HOMAIR) calculated as: [fasting insulin (μIU/ml) × fasting glucose (mmol/l)]/22.5.

rna extraction
Total RNA, including miRNA, was isolated from plasma using Trizol LS according to the manufacturer's protocol (Invitrogen, USA). The concentration and purity of RNA was determined by NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Agarose gel electrophoresis stained with ethidium bromide was used to evaluate the integrity of RNA. All RNA used had A260/280 ratio >1.8 and electrophoresis showed integrity is acceptable.
mirna Microarray expression Profiling 100 ng of total RNA from each plasma sample was labeled and hybridized on human Agilent miRNA microarrays (8*60 K, Design ID: 046064) using the miRNA Labeling Reagent and Hybridization Kit (Agilent Technologies, USA) following the manufacturer's protocol. Briefly, total RNA was dephosphoryl ated, denatured, and then labeled with Cy3CTP. After purifica tion, the labeled RNAs were hybridized onto the microarray. The hybridized array was then washed and scanned using high dynamic range settings according to Agilent specifications and data was extracted from the scanned image using Feature Extraction version 10.2 (Agilent Technologies). The miRNA array data were analyzed for data summarization, normalization, and quality control by GeneSpring GX 12.1 Software (Agilent Technologies). To select the differentially expressed genes, we used threshold values of 2fold change (FC) and a paired ttest P value of 0.05.

selection of neuroendocrine stress response-related candidate mirnas
Among the differently expressed miRNAs from microarray analyze, those involved in biological pathways of stress response, including HAP axis and SNS, were selected for validating their association with T2DM in a larger sample. The criteria of candidate miRNAs, including: (1) the target genes are important regulators of stressinduced activation, including hormone (CRH, COR, N, NE) secreting and hormone receptors; (2) the target genes should be identified by published literature (PubMed database) or at least two prediction programs, including miRecords (2013), TarBase (v 7.0), and miRTarBase (2015).

real-Time Quantitative Pcr
Quantification of miRNAs was performed with a twostep reac tion process: reverse transcription (RT) and PCR. cDNA syn thesis was carried out using the miScript Reverse Transcriptase Kit (Qiagen, Germany) according to manufacturer's instructions with a GeneAmp ® PCR System 9700 (Applied Biosystems, USA). Briefly, each RT reaction contained 1 µg RNA, 4 µl of miScript HiSpec Buffer, 2 µl of Nucleics Mix, and 2 µl of miScript Reverse Transcriptase Mix. The 20 µl reactions were incubated for 60 min at 37°C, 5 min at 95°C, and then held at 4°C. The PCR reactions were performed using 1 µl of the first strand cDNA on LightCycler ® 480 II Realtime PCR Instrument (Roche, Swiss) in total volume of 10 µl with 5 µl of 2 × LightCycler ® 480 SYBR Green I Master (Roche, Swiss), 0.2 µl of universal primer (Qiagen, Germany), 0.2 µl of microRNAspecific primer, and 3.6 µl of nucleasefree water. The reactions were carried out in a 386well plate at 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec, and 60°C for 30 sec. Each sample was analyzed in triplicate.
A melting curve analysis was performed at the end of the PCR cycle to validate the specific generation of the expected PCR product. The miRNAspecific primer sequences were designed in the laboratory and synthesized by Generay Biotech (Generay, PRC) based on the miRNA sequences obtained from the miRBase database (Release 20.0). We used miR451a as an internal control due to its stable and consistent expression throughout all the evaluated samples. Triplicate values of each sample were normalized to miR451a. The miRNA relative expression values were calculated as 2 −ΔCt , [ΔCt = mean Ct (miRX)-mean Ct (miR451a)].

statistical analysis
The data were assessed for normal distribution with the Kolmogorov-Smirnov test and were logarithmically transformed for statistical analyses when necessary. Chisquare test or one way analysis of variance followed by least significant difference test was used to examine the differences between the groups of subjects. Those significant differently expressed miRNAs among the three groups were used in further analysis. The associations between miRNA expression levels and risk of T2DM and IFG were assessed by using both univariate and multivariate logistic regression models with or without adjustment for potential confounders. The relationship between miRNA markers and stress hormone and HOMAIR were evaluated by spearman's correlation coefficient. The miRNA predictors of IR were further identified by stepwise multiple regression. The area under the curve (AUC) of receiver operating characteristic (ROC) was used as diagnostic index for evaluating the biomarker potential of each miRNA and miRNA panel for T2DM. A P value of less than 0.05 was considered statistical significance. All statisti cal analyses were conducted with SPSS software, version 21.0 (IBM SPSS, Inc., Chicago, IL, USA). The graphs were designed by GraphPad Prism 5.01 (Graphpad Software Inc., San Diego, CA, USA).

resUlTs identification of altered mirna expression associated With chronic stress response
The microarray analysis identified 51 human mature miRNAs that differentially expressed in plasma of 10 T2DM patients and 10 controls ( Table 1) by using absolute value of log2FC > 1 and P < 0.05 as cutoff values, including 33 upregulated miRNAs and 18 downregulated miRNAs (Table S1 in Supplementary Material). Hierarchical clustering of miRNA expression patterns correctly classified the two groups (Figure 1). Six miRNAs involved in pathways of chronic stress response, as determined by target gene prediction and literature analysis, were further selected as candidate miRNAs for validation study with a larger sample. These miRNAs are let7b, let7i, miR142, miR144, miR155, and miR29a. The six miRNAs that could significantly discriminate the T2DM from controls and other related details, including the potential target genes were summarized in Table 2.

Basic characteristics of the Participants in the Validation study
The demographic and clinical characteristics of the participants were presented according to the study groups ( Table 3). There was no significant difference in the distribution of age, gender, smoking, alcohol use, and physical activity among three groups (P > 0.05). Most of the clinical parameters, including BMI, WC, SBP, DBP, LDLC, PFG, insulin, HbA1c, and HOMAIR showed significant higher levels in T2DM and IFG groups than those in controls (P < 0.05). Even compared to the IFG group, T2DM group also showed significant higher levels of WC, DBP, LDLC, PFG, insulin, HbA1c, and HOMAIR (P < 0.05).
comparison of stress-related mirna and hormone levels among T2DM, iFg, and control groups The plasma levels of the six stressrelated candidate miRNAs measured by qRTPCR were listed in Table 4. There are four miRNAs, including let7b, miR142, miR144, and miR29a showed significant different expression levels among the three groups (P < 0.05). Compared with control group, expression of let7b, miR144, and miR29a in T2DM group and expression of let7b and miR144 in IFG group were significantly upregulated (P < 0.05) (Figure 2). By contrast, expression of miR142 in subjects with T2DM and IFG were significantly downregulated when compared with control group (P < 0.001). However, there was no significant difference between T2DM and IFG groups for miR29a (P = 0.919) and miR142 (P = 0.064) (Figure 2). As the potential biomarkers of T2DM, the predictive and clinical signifi cance of let7b, miR142, miR144, and miR29a were evaluated in the further statistical analyses.
As presented in Table 4, the plasma levels of E, NE, Cortisol, CRH, and IL6 significantly higher in T2DM group than those in control group (P < 0.05). Cortisol and CRH also showed significantly higher levels in the IFG group when compared with controls (P < 0.05).  The relationship Between mirna expression and stress hormones The spearman's correlation analysis revealed that plasma stress hormones are significantly associated with expression of let7b, miR142, miR144, and miR29a in the study subjects at the multiple test corrected threshold of 0.0125 (assuming family wise a level of 0.05 and number of miRNA markers) ( Table 5). The partial correlation between miRNAs and stress hormones were  further analyzed by correction for FBG and HOMAIR. The sig nificant correlations were consistent with bivariate correlations. However, when data was stratified according to the FBG level, no significant correlation between miRNAs and hormones were identified in the subgroup analysis (P > 0.0125).

association of mirna expression With insulin resistance
Since insulin resistance is the core metabolic abnormality in T2DM, relationships between HOMAIR and expression levels of miRNAs were further assessed. The spearman correlation analysis revealed that let7b, miR144 and miR29a were significant positively related to HOMAIR while miR142 was negatively related to HOMAIR in all subjects ( Table 5). To confirm the association between miRNA and insulin resistance independent of obesity, multiple linear regres sion analysis was further performed with possible confounders were adjusted. As the results indicated, let7b, miR144 and miR29a were significant positive predictors of HOMAIR (P < 0.05) and miR 142 was a significant negative predictor of HOMAIR (P < 0.05) independent of WC ( Table 7). If WC was replaced by BMI among the above covariables in the linear regression model, these miRNAs were also independently associated with HOMAIR ( Table 7).
FigUre 2 | Comparison of plasma microRNAs expression based on quantitative real-time polymerase chain reaction readings among type 2 diabetes mellitus, impaired fasting glucose, and control groups. Data are represented by scatter diagram. *P < 0.05, **P < 0.01, ns, not significant.   Figure 4). These results demonstrated that the miRNA panel had high accuracy in discriminating T2DM from healthy controls (P < 0.001).

DiscUssiOn
This study identified the association between plasma expression of neuroendocrine stress responserelated miRNAs and T2DM as well as IR in an occupational cohort. Based on the differential miRNA expression profile and bioinformatic analysis, six stress related miRNAs (let7b, let7i, miR142, miR144, miR155, and miR29a) associated with T2DM were selected for qRTPCR valida tion. We found that the average expression levels of let7b, miR 142, miR144, and miR29a were significantly different between T2DM patients and healthy controls. The increased expression of let7b, miR144, and miR29a and decreased expression of miR142 were significantindependent predictors of T2DM, IFG, and even IR. At the same time, ROC analysis showed that a three miRNA panel, including let7b, miR142, and miR144 had a high accuracy for diagnosing T2DM.
A number of miRNAs have been shown to be important signa tures or therapeutic targets in patients with T2DM in circulation and specific tissues (25,26). miR184 was reported as a highly regulated miRNA impacting the growth and function of the βcell. By targets Argonaute2, miR184 plays adaptive role of the miRNA pathway based on metabolic state (27). Stressinduced overexpression of miR708 suppressed βcell proliferation and induced βcell apoptosis, which provide a novel mechanism of glucose regulation of βcell function and growth (28). Comparing to tissue biomarkers, circulating miRNAs are preferred due to easier sampling and testing. Circulating miR192 and miR193b were even found significantly increased in the prediabetic state (29). Some other miRNAs have also been identified as biomarkers for diabetic complications, such as plasma miR152 and diabetic nephropathy, and miR93 and diabetic retinopathy (30,31). Therefore, circulating miRNAs may help to predict the develop ment and progression of T2DM. Increasing evidence indicates that psychosocial factors facilitate the development and progression of central obesity and T2DM, possibly as the results of increasing insulin resistance (32). In our previous study, the Copenhagen Psychosocial Questionnaire was used to assess jobrelated psychosocial stress. We found that after adjustment for the influence of cortisol, HOMAIR was also significantly associated with scales of "demands at work" and "insecurity at work" (6). Psychosocial factors interact with biological mechanisms at several levels in the organisms include neuroendocrine, metabolic, and molecular responses. Stress hormones influence glucose metabolism by stimulating gluco neogenesis and impairing glucose uptake in peripheral tissues like fat and muscle (8,33). Activation of stress hormones secre tion is mediated by neuronal signals. Repeated stress activates hypothalamic paraventricular nucleus to synthesize CRH, the major regulator of stress responses. Anterior pituitary responds to CRH to produce adrenocorticotropic hormone, which is released into the circulation and promotes synthesis and release of glucocorticoids (cortisol in humans) at the level of the adrenal gland (34). The HPA axis controlling cortisol secretion and the SNS both originate in the hypothalamus and they are interlinked, and hyperactivity in one of them can activate the other (35).
Our study showed that plasma levels of E, NE, cortisol, and CRH significantly higher in T2DM group than those in control group (P < 0.05). Cortisol and CRH also showed significantly higher levels in the IFG group when compared with controls (P < 0.05). These results provide evidence for the association between chronic stress and T2DM based on an occupational popu lation study. Elevated activities in both HPA axis and SNS, which were triggered by psychosocial factors have also been reported in individuals with the metabolic syndrome (36). Increased plasma IL6 levels in T2DM patients and correlation of IL6 with cortisol in this study maybe another evidence in the pathway between chronic stress and T2DM, since various psychological stressors alone can induce proinflammatory (36,37). Cellular responsiveness to stress hormone depends on the amount of corresponding receptor protein. Significant upregulated expression of let7b and miR144 in T2DM patients in our study may indicates downregulated expression of their target gene, including ADBR2/ADBR3 (βadrenoreceptors) and NR3C1 (GR), respectively. Sheng, et al. verified that the protein expression of ADRB3 was regulated by direct binding of let7b at 3′UTR (22). In addition, they found that let7b was negatively correlated with ADRB3 protein level and had a decreased expression in perivas cular adipose tissue of hypertensive mice. Cortisol treatment reduced NR3C1 mRNA levels and increased miR144 expression in hippocampus and sperm of mice (24). The evidence of NR3C1 as the target of miR144 can also be obtained from nextgeneration sequencing (miRTarBase). Positive correlation between let7b and E/NE, miR144 and cortisol in the study subjects indicated decreased stress hormone receptor expression, which reflects the negative feedback mechanisms for controlling the magnitude of responses to stress (34). Our previous populationbased study showed that chronic stress increase plasma cortisol, which subse quently decrease mRNA expression of GRα and increase of GRβ/ GRα mRNA ratio in lymphocyte (14). miR142, which mediated the expression of CRH, showed lower levels in either T2DM or IFG individuals compared with controls. miR142 also participates in the inflammatory response by regulating the production of IL6 (23). Both plasma CRH and IL6 were negatively associated with miR142 expression in total subjects, which may be explained by the inhibitory effect of miRNA on target genes. Microarray analysis revealed the down regulation miR142 in the peripheral blood of mice associated with pronounced enhanced oxidative stress (22). miR29a was reported to be involved in the glucocorticoids signaling pathway (24) and chronic academic stress excessive glucocorticoids reduce the expression of miR29a (13). The association may partially be explained by receptor gene NR3C1 and CRHR1. Although miR29a expression significantly associ ated with T2DM and HOMAIR, relatively weak correlations exist only between miR29a and cortisol (r = 0.183) in the total subjects. It is possible that miR29a regulates glucose metabolism though other pathway.
It is the first time that we systematically investigated the relationship between plasma expression of neuroendocrine stress responserelated miRNAs and T2DM at population level. The reasonable diagnostic accuracy of miRNA panel (let7b, miR142, and miR144) indicates their clinical value in diagnosing T2DM. It is worth noting that significantly altered expression of let7b, miR142, miR144, and miR29a was found even at the stage of prediabetes in this study. These miRNAs were also identified as significantindependent predictors of insulin resistance which is an important component of the pathophysiological processes that underlies the development of T2DM. The findings of this study support the hypothesis that neuroendocrine stress response related miRNAs may play a role in the pathogenesis of T2DM and insulin resistance by regulating those target genes which involve neuroendocrine pathways of stress responses, which unveils new targets for the prevention, prediction, and more personalized treat ment of neuroendocrine stressrelated disorders, including T2DM.
Our results are reliable based on the threestage design. However, there are several limitations in our study. The asso ciation between neuroendocrine stress responserelated miRNAs and T2DM could not be proved causality; followup studies are needed to confirm our findings. Experimental study is needed to clarify the role of these miRNAs in the molecular mechanisms of T2DM.

eThics sTaTeMenT
This study was approved by the Ethical Committee of Capital Medical University.
aUThOr cOnTriBUTiOns YXY designed the study. YZL and JD collected the data. YXY and YZL conducted bioinformatical analyses and statistical analyses. YZL, JZ, SW, and YH conducted the experiments. All authors interpreted the data and all authors contributed to writ ing. All authors have approved the final manuscript.

FUnDing
This study was supported by the National Natural Science Foundation (81573214, 81773511), the Beijing Municipal Natural Science Foundation (7162020), the Scientific Research Project of Beijing Municipal Educational Committee (KM201510025006).