Modified level of miR‐376a is associated with Parkinson's disease

Abstract Parkinson's disease (PD) is a frequent progressive neurodegenerative disorder. Impaired mitochondrial function is a major feature of sporadic PD. Some susceptibility or causative genes detected in PD are strongly associated with mitochondrial dysfunction including PGC1α, TFAM and GSK3β. microRNAs (miRNAs) are non‐coding RNAs whose altered levels are proven in disparate PD models and human brains. Therefore, the aim of this study was to detect modulations of miRs upstream of PGC1α, TFAM and GSK3β in association with PD onset and progress. In this study, a total of 33 PD subjects and 25 healthy volunteers were recruited. Candidate miRNA (miR‐376a) was selected through target prediction tools and literature survey. Chronic and acute in vitro PD models were created by MPP+‐intoxicated SHSY5Y cells. The levels of miR‐376a and aforementioned genes were assessed by RT‐qPCR. The expression of target genes was decreased in chronic model while there were dramatically up‐regulated levels of those genes in acute model of PD. miR‐376a was strongly altered in both acute and chronic PD models as well as PBMCs of PD patients. Our results also showed overexpression of PGC1α, and TFAM in PBMCs is inversely correlated with down‐regulation of miR‐376a, suggesting that miR‐376a possibly has an impact on PD pathogenesis through regulation of these genes which are involved in mitochondrial function. miR‐376a expression in PD‐derived PBMCs was also correlated with disease severity and may serve as a potential biomarker for PD diagnosis. This is the first study showing altered levels of miR‐376a in PD models and PBMCs, suggesting the probable role of this miRNA in PD pathogenesis. The present study also proposed TFAM and PGC1α as target genes of miR‐376a for the first time, through which it possibly can exert its impact on PD pathogenesis.


| INTRODUC TI ON
Parkinson's disease (PD) is the second most frequent neurodegenerative disorder (NDD) whose pathological characteristics are progressive loss of dopaminergic neurons of midbrain and presence of intraneuronal cytoplasmic inclusions, namely "Lewy bodies". 1,2 Around 95% of PD cases are sporadic and familial forms constitute the remainder. 3 While the precise pathomechanism of PD is not entirely understood, some molecular mechanisms are suggested to contribute to the pathogenesis, including chronic neuroinflammation, oxidative stress, apoptosis, autophagy and mitochondrial dysfunction. 1,3,4 There is no effective treatment approach to halt the development of PD currently, and its management remains symptomatic. Recently, substantial efforts have been made to comprehend molecular mechanisms responsible for PD in order to develop new therapeutic strategies for the disease. 3,5 Considering that mitochondrial dysfunction is a prominent pathological hallmark of many neurodegenerative disorders, understanding the mechanisms through which mitochondrial dysfunction takes part in the pathogenesis of these diseases can help to define novel therapeutic approaches. 6 Thus, complex 1 inhibitors like 1-methyl-4-phenylpyridinium (MPP + ) are extensively applied as a model for mitochondrial dysfunction. 7,8 The mechanism of MPP + neurotoxicity involves inducing oxidative stress, mitochondrial dysfunction and inflammation which finally lead to apoptotic cell death. 4 Here, we served MPP + as a means to reproduce PD-related mitochondrial dysfunction in SH-SY5Y cell line.
Notably, environmental factors as well as genetic susceptibility are decisive factors in sporadic PD aetiology. 9 Beside environmental agents, several susceptibility or causative genes were detected in PD, some of which are strongly associated with mitochondrial dysfunction including peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC1α), mitochondrial transcription factor A (TFAM) and glycogen synthase kinase 3 beta (GSK3β). 1 PGC1α is a multifunctional coactivator of transcription factors like NRF-1, NRF-2, TFAM and FOXO receptors, and thereby regulates mitochondrial biogenesis, respiration as well as function. 1,3 Moreover, since PGC1α induces the expression of several reactive oxygen species (ROS) detoxifying enzymes including SOD1 and 2, catalase and glutathione peroxidase-1, it tightly regulates oxidative capacity and reduces oxidative stress. 3,10 There is a substantial literature indicating possible links between PGC1α and diverse neurodegenerative diseases (NDDs), comprising amyotrophic lateral sclerosis (ALS), Huntington's disease (HD), Alzheimer's disease (AD) and PD. 10 As a target gene of PGC1α, TFAM regulates mitochondrial genome replication and transcription as well as mitochondrial biogenesis. 11 TFAM also makes a positive contribution to mtDNA stability and initiates synthesis of mtDNA-encoded respiratory chain subunits. 7 GSK3β is widely known for its role in the pathogenesis of a host of neurodegenerative disorders comprising PD, impacting several pathological processes such as neuronal apoptosis, neuroinflammation and dopaminergic (DA) neuron degeneration. 12 miRNAs, a class of non-coding RNAs, are roughly 22 bases in length and take part in a wide variety of biological and pathological processes in health and disease. 5,13 They tend to repress genes' expression level through binding to specific sequences mostly located in the 3'-untranslated region (3'UTR) of mRNAs, culminating in mRNA degradation or deadenylation and translation inhibition. 5,14 miRNAs have an enormous modulatory potential considering that each single miRNA can govern expression of hundreds of different transcripts and that's why whole phenotype of a disease can be feasibly affected by regulating a single miRNA molecule. 5,15,16 Mounting evidence has emerged showing the association between miRNA network and pathogenesis of neurodegenerative disorders including PD. 12 Furthermore, altered level of several miRNAs has been reported in disparate PD models and human brains of patients suffering from PD. 5,12,13,15,16 While there is substantial literature confirming miRNAs' implication in PD development, only a few studies have attempted to identify deregulated miRNAs. 5,16 Since maintenance of mitochondrial machinery depends on the action of some mitochondria-related nuclear-encoded proteins, 11 the principal aim of this study was to assess the expression levels PGC1α, TFAM and GSK3β, as well as their predicted upstream miRNA, miR-376a, in chronic and acute models of PD and patient's peripheral blood mononuclear cells (PBMCs).

| Sample collection and PBMC isolation
Our study design focused on the evaluation of disparities between PD patients and clinically healthy subjects in terms of gene expression. Therefore, a total of 33 PD subjects were recruited, attending the Parkinson's disease Assessment Unit of the Alzahra Hospital (Isfahan, Iran). The protocol of study to use human samples was confirmed by both the Bioethics Committee of University lateral involvement only usually with minimal or no functional disability (HY-1 stage) and fourteen patients with bilateral or midline involvement without impairment of balance (HY-2 stage) were categorized as the early-stage PD. However, four patients suffering from mild-to-moderate disability with impaired postural reflexes but physically independent (HY-3 stage) and three patients with severe disability being able to walk or stand independently without the use of an assistive device (HY-4 stage) were regarded as the advanced stage PD.
All participators in the study signed a written informed consent. Approximately four millilitres of venous blood was collected from all subjects. Within two hours of blood collection, PBMCs were separated from whole blood using Ficoll-Histopaque density gradient centrifugation according to the manufacturer's instruction.

| Candidate miRNA selection
According to recent studies, three mitochondria-related genes, including PGC1α, TFAM and GSK3β whose expressions were significantly altered in different neurodegenerative conditions, were taken into account to select a putative targeting miRNAs. TargetScan
Culture medium was exchanged every 3-4 days, and the cells were harvested and dispersed at 65%-75% confluency. MPP + (Sigma, USA) was diluted in medium with low serum to four ultimate concentrations. Twenty-four hours prior to initiating each experiment, culture medium was changed to 1% FBS. To create an acute toxicity cellular model, SH-SY5Y cells were exposed to 500, 1000, 2000 and 3000 μmol/L MPP + for 24 hours. Chronic model of toxicity was also produced by MPP + treatment with the same concentrations three times per week for 2 weeks.

| Flow cytometric detection of apoptotic cells
Apoptosis was detected by cytofluorometric analysis using a FACSCalibur (Becton Dickinson, Mountain View, CA, USA).
Apoptotic rate induced by chronic or acute MPP + toxicity was quantified using the Phosphatidyl Serine Detection Kit (IQ products-) based on the manufacturer's recommendations. Briefly, after exposure to MPP + , SHSY-5Y cells were collected, washed and incubated with 10 μL of Annexin V-FITC at 4°C in the dark for 20 minutes and then, 10 μL of PI. Ultimately, the fluorescence intensity was analysed by FACSCalibur flow cytometer and CellQuest software.

| Intracellular ROS measurement
The amount of ROS was estimated using dichlorofluorescein-diac-

| Statistical analysis
All statistical analyses were conducted by SPSS 22.0. and GraphPad Prism 6 software. The disparity between means was analysed by Mann-Whitney test, Student's independent-samples t test, chi-square test as well as one-way ANOVA. The Kolmogorov-Smirnov normality test (KS-test) was served to evaluate the normality of data distribution. To examine the diagnostic sensitivity and specificity of miR-376a level to discriminate between PD cases and healthy subjects, the receiver operating characteristic (ROC) curve was drawn and MEDCALC (www.medca lc.org) online software was also recruited to calculate positive predictive value (PPV) and negative predictive value (NPV).
Spearman coefficient calculation was also used to analyse possible correlations between parameters of interest. Data are expressed as the mean ± SD of three independent experiments, and P-values <.05 were deemed to be significant differences statistically.

| Characteristics of study participants
Demographic information and clinical information of all subjects are listed in Table 2. Although statistical analyses demonstrated no significant differences in terms of gender and age between two groups of subjects and stages, there was a significant correlation between PD stages and disease duration (P = .007).

| In silico findings
To choose a candidate miRNA, following criteria were considered: (a) strong predicted interactions between the miRNA and candidate genes.  Note: Data are presented as mean ± SD. Two diagnostic groups and stages had no significant differences in distributions of age and sexuality, while there was a significant correlation between PD stages and disease duration (P = .007).
Abbreviations: F, female; M, male; PD, Parkinson's disease; HY, Hoehn and Yahr scales. a P-values were calculated using chi-square test.
b P-value was calculated by Student's t test.
c P-value was calculated using one-way ANOVA.

| Dose-dependent cell viability loss after MPP + treatment
To evaluate the survival of SHSY-5Y cells in response to acute oxidative damage, the cells were exposed to various doses of MPP + Optimal MPP + concentration to create chronic PD model was also selected based on MTS assay. SHSY-5Y cells were exposed to the same concentrations of MPP + three times per week for up to 2 weeks. According to MTS analysis, 1000 μmol/L MPP + which induced about 37% cell death was chosen as an appropriate concentration to make a chronic cellular model of PD (Figure 2A).

| MPP + -induced increase in apoptosis
Rate of apoptotic cell death induced by MPP + was quantified by Annexin

| MPP + -induced ROS overproduction
Intracellular ROS formation was assessed by detecting alterations in DCF fluorescence intensity using flow cytometry. To examine MPP + effect on ROS generation and subsequent oxidative stress, SHSY-5Y cells were exposed to MPP + acutely and chronically for 24 hours and 2 weeks, respectively. In the control group of acute toxicity, the pro-

| Down-regulation of miR-376a and upregulation of genes' transcript levels after acute MPP + treatment
In order to evaluate alterations in candidate genes' expression, RT-qPCR was performed, representing that TFAM, GSK3β and particularly, PGC1α levels were significantly affected by acute oxidative damage.

TA B L E 3 (Continued)
The transcript levels of PGC1α markedly raised following treatment with 2000 μmol/L MPP + for 24 hours, in comparison with that of unexposed control. Similarly, TFAM and GSK3β were substantially up-regulated following exposure to MPP + ( Figure 3A). The expression levels of candidate genes were normalized to GAPDH as a reference gene. Alteration in miR-376a level was also quantified by real-time PCR analysis, and expression was normalized to U6 snRNA that has been found to be an appropriate internal control for miRNAs' studies in SHSY-5Y cells. Exposing SHSY-5Y cells to 2000 μmol/L MPP + for 24 hours culminated in a dramatic drop in miR-376a level in comparison with untreated cells as a control group ( Figure 3B).

| Up-regulation of miR-376a and downregulation of genes' mRNA levels in response to chronic MPP + treatment
In spite of acute MPP + model, levels of PGC1α, TFAM and GSK3β transcripts dropped dramatically in SHSY-5Y cells chronically stressed with MPP + ( Figure 3C). Moreover, low dose, repeated 1000 μmol/L MPP + administration caused a significant growth in miR-376a level contrary to acute toxicity ( Figure 3D).

| Up-regulation of miR-376a and downregulation of genes' mRNA levels in PBMCs derived from PD patients
PBMCs of 33 PD subjects and 25 age-matched controls were assessed by measuring the mRNA levels of three mitochondria-related genes. As depicted in Figure 4A, A statistically dramatic drop in both PGC1α and TFAM levels occurred in PD patients compared with agematched controls. Additionally, PD PBMCs showed a marked decline in GSK3β mRNA expression compared with healthy controls. Our results also represented that miR-376a was strongly up-regulated in PD patient's PBMCs ( Figure 4B).

| miR-376a expression in PD PBMCs is correlated with disease severity
In order to gain more insight into the connection between the progression of PD and miR-376a level, statistical analysis was performed showing that the expression of miR-376a in advanced stages is significantly higher than that in stages 1 and 2 of PD and increases with disease severity ( Figure 4C).

| Diagnostic significance of miR-3 6a expression
The suitability of miR-376a to distinguish PD patients from healthy controls was investigated by performing ROC analysis. Higher total area under the curve (AUC) value demonstrates more effective overall performance of the diagnostic marker to accurately discriminate between certain two conditions. Hence, total are under the ROC curve represents the fact that miR-376a level (AUC = 0.8024, P < .0001. 95% CI = 0.6878 to 0.9171) may serve as a biomarker for discriminating between PD cases and normal controls ( Figure 5A). Notably, miR-376a also had relatively high NPV (72.41%, 95% CI = 58.37% to 83.09%) and PPV (86.21%, 95% CI = 71.38% to 94.00%), showing diagnostic value of this miRNA in PD.

| Decreased expression of PGC1α and TFAM negatively correlates with miR-376a expression in PD PBMCs
According to opposing patterns of miR-376a and three candidate genes in cellular models, we hypothesized that altered miR-376a  PGC1α is a master regulator of mammalian mitochondrial biogenesis in response to disparate physiological or pathological stresses and deemed to be a valuable potential therapeutic target for PD. 9,10 PGC1α plays an indispensable role in normal function of mitochondria, and its deficiency can participate in neurodegenerative processes. On top of that, its overexpression was proven to be neuroprotective against neurotoxins in several models. 1 Numerous studies demonstrated that PGC1α expression can be induced in neurons exposed to oxidative stress. 4 Totally, studies on changes in PGC1α levels have provided inconsistent data in different PD F I G U R E 2 MPP + -induced cell death, apoptosis and intracellular ROS overproduction in SHSY-5Y cells. (A) SHSY-5Y cells were exposed to various concentrations of MPP + for 24 h and 2 wk. Cell viability was monitored using MTS assay. Two thousand and 1000 μmol/L MPP + were selected as optimal concentrations for creating acute and chronic PD models because of inducing cell death by 42% and 37%, respectively. levels in PD PBMCs. 28  cells. 31 Using PBMC as a model of PD, we examined if peripheral expression of miR-376a is altered in this condition. As miR-376a is one of the most highly expressed miRNAs in elderly human blood mononuclear cells, aberrant levels of this miRNA may strongly affect its target genes and molecular networks normally controlled by this miRNA in PBMCs. 32 Therefore, we aimed here to determine miR-376a alteration in PBMCs from patients with PD compared to healthy subject and found profound increase in miR-376a expression in PD PBMC samples relative to healthy controls. Upregulated level of miR-376a was also detected in T cells of MS patients. 13 miR-376a has been identified to be down-regulated in the brains of late-onset form of Alzheimer's disease (LOAD). 14 Altered abundance of miR-376a was also reported in other neurodegenerative conditions such as spinocerebellar ataxia type 1 (SCA1) and prion disease. 13,33 When correlating the miRNA expression with the indices of disorder severity, interestingly, we found out that miR-376a expression was significantly higher in stage III/IV PD patients than in stage I/II patients and was strongly correlated with disease severity.

| D ISCUSS I ON
Evaluating the possible association of miR-376a with predicted target genes, PGC1α, TFAM and GSK3β, was the following stage to proceed in the current study. Opposing expression pattern of the miR-376a with the genes in 3 assessed neurodegenerative conditions was regarded as a validating evidence of predicted interactions.
In addition, the significance of the associations between miR-376a and PGC1α and TFAM was further validated by Spearman's correlation coefficient data analysis (r = −.5321 and P = .0014, r = −.5424 and P = .0011, respectively), meaning that miR-376a may affect mitochondrial function in PBMCs by modulating PGC1α and TFAM expression.
To evaluate the overall modulation of miR-376a in discriminating between PD patients and controls, sensitivity, specificity, NPV and PPV were calculated. Accordingly, AUC in ROC analysis revealed high sensitivity and specificity of miR-376a in terms of PD patients' distinction from healthy individuals, demonstrating the possible involvement of miR-376a in PD, while the mechanisms by which miR-376a may contribute to PD development await to be precisely identified. Notably, single miRNA expression profile change might not demonstrate robust PD classifiers across various PD subject cohorts, while combinatory miRNA signatures represent more promising complementary diagnostic tools for this disorder. Thus, it is imperative that our future studies evaluate the diagnostic potential of miR-376a accompanied by other candidates as a miR combination acting as a good classifier across different patient cohort studies. 34 Collectively, the same expression patterns of miR-376a and genes in chronic model and PD PBMCs suggest that possibly lowdose chronic toxicity model can more closely exhibit chronically developing PD than high-dose acute MPP + models. 35 The current study also revealed that acute severe mitochondrial dysfunction in SHSY-5Y cells induced protection via enhancing the expression of PGC1α, TFAM and GSK3β. However, low-dose chronic MPP + treatment did not induce those mechanisms and instead suppressed expression of genes. Similarly, the expression levels of PGC1α, TFAM and GSK3β decreased in PBMCs of PD patients indicating that brain mitochondrial dysfunction is also reflected in PD PBMCs. Given that miR-376a and possible target genes showed opposite expression patterns in 3 assessed neurodegenerative conditions and negative correlations were observed between miR-376a and mRNA levels in PBMCs, we can suggest that miR-376a may be implicated in the pathogenesis of PD possibly by regulating expression of mitochondria-related genes.
However, further studies have to be performed to better understand the mechanism enabling this miRNA to operate on its targets in PD. Notably, our findings revealed that overexpression of miR-376a in PBMCs might be considered as a potential marker of the diagnosis of PD. However, larger-scale studies using brain samples from PD patients along with independent additional cohort studies must be conducted to confirm the diagnostic significance of this miR.

ACK N OWLED G EM ENTS
We would like to show our gratitude to our colleagues especially Parkinson's Research Team (PRT) at Royan Institute for Biotechnology for their assistance and helpful comments.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest. All authors support submission to this journal.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.