Alterations in the Expression Levels of miR-9-5p and miR-4467 in Peripheral Blood of Patients with First-Episode Schizophrenia

The diagnosis of schizophrenia (SCZ) patients largely rely on abnormal behavior and mental state examination, which is inevitably subjective and susceptible. Finding more stable and objective biomarkers has always been the vision of scholars. In this study, we aim to validate peripheral blood miRNA as potential biomarkers for patients with rst-episode schizophrenia (FES). We performed high-throughput sequencing analysis and screened 138 differentially expressed miRNAs (DEmiRNAs) in the peripheral blood in a test cohort (15 FES patients and 15 healthy controls), followed by quantitative reverse transcription-polymerase chain reaction (qRT-PCR) in miR-9-5p, miR-144-3p, miR-328-3p, and miR-4467 in a validation cohort (35 FES patients and 60 healthy controls), compared with healthy controls, miR-9-5p of FES patients was signicantly down-regulated, and miR-4467 was up-regulated. The receiver operating characteristic curve (ROC) demonstrated that miR-4467 had a better diagnostic value than miR-9-5p. Bioinformatics analysis predicted the potential target genes of miR-9-5p and miR-4467 mainly involved in the regulation of body behavior, neuronal differentiation, and nervous system development, and enriched in Neurotrophin, MAPK, and phosphatidylinositol signaling systems. The protein interaction assay identied NEDD4, EIF4G1, FBXL16, FBXL3, etc. as the hub target genes. Summarily, our data suggested the expression of miR-9-5p and miR-4467 in peripheral blood might serve as promising biomarkers for FES patients, and might involve in the occurrence and development of schizophrenia by regulating gene modules related to neurodevelopment and neuroprotective. - review &


Introduction
Schizophrenia affects approximately 1% of the population in the world and is closely related to human disability, mortality, and economic burden on individuals and society (Baron 2002). Currently, the diagnosis of schizophrenia is mainly based on abnormal behavior and mental state examination by psychiatrists, which is inevitably subjective and susceptible. In view of the fact that early diagnosis and intervention of schizophrenia have a signi cantly positive impact on the prognosis, there is still a great need for detectable biomarkers to assist in the diagnosis of SCZ.
MicroRNAs (miRNA, miR) are a class of non-coding small RNA, which can regulate up to one-third of human genes by incomplete base pairing with the target 3'-UTR sequence at the post-transcription level (Krol et al. 2010). And over 60% of human protein-coding genes are regulated by miRNA (Sayed and Abdellatif 2011). Alteration in miRNA expression has been con rmed in a variety of human diseases,  Perkins et al. 2007), but the invasiveness of sampling made it impossible to apply to living patients. Therefore, developing miRNAs as non-invasive biomarkers for blood diagnosis and treatment effect evaluation has become a popular eld of interest for researchers.
There have already been ndings of available miRNAs useful in differentiation of SCZ patients. Hui Wei et al. discovered dysregulation of plasma miR-130b and miR-193a-3p could be biomarkers for SCZ (Wei et al. 2015). He et al. identi ed that miR-34a-5p and miR-432a-5p expressed aberrantly in the serum of SCZ patients (He et al. 2019). E Gardiner et al. performed microarray analysis and qRT-PCR veri ed seven downregulated miRNAs in PBMCs of schizophrenia patients (Gardiner et al. 2012). However, miRNAs with altered expression in the initial stage of schizophrenia have not been fully investigated. We believe that these miRNAs are of vital importance to the understanding of the pathogenesis of SCZ and the early diagnosis. Therefore, we designed a case-control study to screen abnormally expressed miRNAs in peripheral blood of rst-episode schizophrenia (FES) patients by high-throughput sequencing. Then performed the qRT-PCR analysis in an expanded sample to con rm the initial nding. We also used bioinformatics methods to predict potential target genes and integrate biomarkers information to further understand the possible biological pathways and regulatory mechanisms of the development of schizophrenia.

Participants
35 FES patients admitted to Changchun Psychological Hospital from July 2017 to August 2019 were recruited in this study. All patients were diagnosed by at least two consultant psychiatrists according to the Tenth Revision of International Classi cation of Diseases (ICD-10) for SCZ and were recruited before any antipsychotic medication. Patients with severe medical illnesses, other neuropsychiatric disorders, structural brain disorders, brain trauma, previous continuous illegal substance use, or movement disorders were excluded. In addition, 60 healthy individuals with no psychiatric disorders or family history of mental illnesses were recruited as the control group.
Among them, high-throughput sequencing samples comprised 15 SCZ patients and 15 controls matched on age and gender as the rst set. On the basis of the sequencing samples, according to the above inclusion and exclusion criteria, we continue to recruit individuals with FES to 35 cases, and healthy controls to 60 cases as the second set for veri cation. This study was approved by the Ethics Committee of the School of Public Health, Jilin University. Informed consent of the participants was obtained after the nature of the procedures had been fully explained.
2.2 Peripheral blood collection and total RNA extraction 1.5ml whole blood was obtained from all fasting subjects, treated with 4.5ml Trizol reagent (TaKaRa,Japan), and put into a liquid nitrogen tank for quick freezing. When extracting total RNA, transfer the mixed solution to four 1.5ml EP tubes, let the sample stand for 5 minutes at room temperature, and centrifuge at 12,000 rpm at 4°C for 5 minutes. Collect the supernatant, add 0.2ml of chloroform and shake well, then let it stand at room temperature for 5 minutes, and centrifuge at 12000rpm at 4°C for 15 minutes. Take the supernatant and add 0.8ml of isopropanol reagent to mix them well. After standing for 10 minutes at -20°C, centrifuge at 12000rpm at 4°C for 10 minutes. Subsequently, washed the total RNA pellet with 75% ethanol and centrifuge at 12,000 rpm at 4°C for 10 minutes. The supernatant was completely discarded, and the RNA pellet was dried at room temperature. Finally, the total RNA was dissolved in 15uL of DEPC and stored at -80°C.

High-throughput sequencing analysis
Sequencing was completed by Beijing Novogene Experimental Department. Brie y, after the samples were quali ed, we used the Small RNA Sample Pre Kit to construct a library, and total RNA was used as the starting sample to synthesize cDNA by reverse transcription, PAGE gel electrophoresis was used to separate the target DNA fragments, and we cut the gel to recover the cDNA library. After the library was constructed, use Qubit2.0 for preliminary quanti cation, dilute the cDNA to 1ng/ul, and then use Agilent 2100 to detect the insert size of the library, and use qPCR to accurately quantify the effective concentration of cDNA (effective concentration > 2nM). The library preparations were then sequenced on the Illumina HiSeq 2500 platform. DNA polymerase, adaptor primers, base-speci c uorescently labeled dNTP, and buffer were added to the reaction system, and the laser was used to excite the uorescent signal, and nally computer analysis was conducted to convert the optical signal into sequencing bases. miRNA expression levels were estimated by read counts, differential expression analysis of two samples was performed using the edgeR package. |log 2 Foldchange|>0.5 and p < 0.01was set as the threshold for signi cantly differential expression by default.

qRT-PCR analysis
After checking the concentration and purity of RNA,4 µLtotal RNA was reverse transcribed to cDNA using TransScript miRNA First-Strang cDNA Synthesis SuperMix (TransGen Biotech, Beijing, China). The reverse transcription reactants were mixed according to the instruction and the reaction was performed in Mastercycler nexus PCR instrument at 37°C for 1h, and RT Enzyme Mix inactivated at 85°C for 5s. Afterward, PCR reactions were performed in a 10uL reaction mixture containing 5µL 2*TransStart® Tip Green qPCR SuperMix (TransGen Biotech, Beijing, China), 0.6µL each of forwarding primer and universal primer (Sangon Biotech Shanghai China, see the primer sequences in Table 1), and 2µL diluted RT cDNA product. All reactions were run in triplicate. PCR was performed on LightCycler®96 (Roche) at 94℃ for 30 seconds followed by 44 cycles at 94℃ for 5 seconds, 64℃ for 15 seconds, and 72℃ for 10 seconds. U6 was selected as the internal reference gene for normalization, and the relative expression levels of miRNAs were calculated by the 2 −△△Ct method.

Bioinformatics Analysis
Bioinformatics methods were used to predict potential target genes of miRNAs with differential expression in SCZ patients and controls. We rst take the intersection of the prediction results of TargetScan (http://www.targetscan.org/vert_72/) and miRDB (http://mirdb.org/), and then take the union set withresults from miRTarbase (http://mirtarbase.cuhk.edu.cn/php/index.php) as the target gene set of miRNAs. We used DAVID (https://david.ncifcrf.gov/home.jsp) to perform GO and KEGG pathway enrichment analysis on the target genes, and use FDR (false discovery rate) corrected Padj < 0.05 as standard to screen and sort out biological functions and signaling pathways which are signi cantly related to target genes. The protein and protein interaction networks were constructed by String (http://www.string-db.org), and Hub genes were screened by Cytoscape MCC algorithm.

Statistical analysis
For the high-throughput sequencing analysis in the rst set, the statistical signi cance of readcount was assessed using Student's t-test. Mann-Whitney U test and Chi-square test were used to compare continuous and categorical variables of basic demographic data between groups, respectively. One sample Kolmogorov-Smirnov test was used to test the normal distribution of variables. The relative expression of miRNA was expressed by the median and quartile of 2 −△△Ct , Mann-Whitney U test was used to compare the expression of miRNAs between groups. ROC curve was drawn to assess the diagnostic value of miRNAs for schizophrenia. G*Power 3.1.9.2 was used to calculate the minimum sample size. Statistical analysis and graphs plotting were completed using SPSS 24.0 and GraphPad Prism 8.0.1. p < 0.05 (two-tailed) was considered statistically signi cant.

Characteristics of subjects
There was no statistical difference between the FES group and the control group in gender, age, years of education, and residence. However, the family history of mental illness (p = 0.002) and marital status (p < 0.001) were signi cantly different between the two groups ( Table 2).  In the veri cation part, compared with healthy controls, the expression of miR-9-5p was signi cantly down-regulated in the peripheral blood of patients with schizophrenia (p = 0.046), and miR-4467 was signi cantly up-regulated (p < 0.001), which was consistent with high-throughput sequencing analysis. However, no signi cant difference in the expression of miR-144-3p and miR-328-3p between the two groups was found in the expanded sample (Table 4), which were excluded for subsequent analysis. To further assess the diagnostic value of miRNAs, ROC curves were plotted, and AUC was calculated to evaluate the predictive power of these two miRNAs for schizophrenia. The results showed that miR-4467 (AUC = 0.719, 95%CI: 0.613-0.824) showed a greater diagnostic value than miR-9-5p (AUC = 0.623, 95%CI: 0.509-0.737) (Fig. 2). 3.4 Target gene prediction and functional enrichment of miR-9-5p and miR-4467 Taken the intersection and union of the results as planned, 1116 target genes for miR-9-5p and 92 target genes for miR-4467 were predicted using Targetscan, miRDB, and miRTarbase. Figure 3 showed the biological processes (BP), cell components (CC) and molecular functions (MF) the target genes of miR-9-5p and miR-4467 mainly involved in, including locomotory behavior, nervous system development, and regulation of cardiac muscle contraction. regulation of neuron differentiation and development, negative regulation of Schwann cell proliferation; KEGG pathway analysis showed that the potential functions of miR-9-5p target genes are mainly enriched in neurotrophin, MAPK, Estrogen, and Phosphatidylinositol signaling systems (Fig. 4).

Discussion
In recent years, the view that miRNAs participate in the pathophysiological process of mental diseases by regulating hundreds of target transcripts has been widely accepted. Evidence showed that the dysregulation of many miRNAs in postmortem brain samples was associated with schizophrenia Therefore, this study recruited FES patients who were diagnosed for the rst time and had not been treated with antipsychotic drugs, analyzed the alteration of miRNA expression pro les in their peripheral blood, predicted target mRNAs and integrated their common functions. We initially screened the differentially expressed miRNAs in schizophrenia patients through high-throughput sequencing globally, subsequently performed a qRT-PCR procedure on 4 miRNAs (miR-9-5p, miR-328, miR-144, and miR-4467) in an expanded sample for veri cation to avoid false-positive results. We found that compared with healthy controls, the level of miR-9-5p in peripheral blood of FES patients was down-regulated, and miR-4467 was up-regulated. Moreover, the ROC curve indicated that, as a non-invasive biomarker of FES, the expression level of miR-9-5p in peripheral blood showed a better diagnostic value than miR-4467.
There have been reported cases earlier in researches focusing on the two miRNAs and mental diseases. miR-4467 was rst found in the cerebrospinal uid of Alzheimer's patients and was identi ed as a potential biomarker for Alzheimer (Denk et al. 2015). Disorders of the brain can exhibit similar symptoms (Anttila et al. 2018), which made us wonder if the etiology of schizophrenia and Alzheimer would overlap. It is worth mentioning that this study is the rst attempt that the alteration of miR-4467 expression was explored in SCZ. As for miR-9-5p, Mehmet et al. In order to better understand the role of miR-9-5p and miR-4467 in schizophrenia, we performed bioinformatics analysis to predict their target genes. In addition to the top 10 functions of potential target genes shown in Fig. 3, the SCZ-related biological processes that miR-9-5p may participate in also include regulation of astrocyte, oligodendrocyte differentiation, regulation of neural precursor cell and mesenchymal cell proliferation, regulation of neuron differentiation, development, and apoptotic, the biosynthetic process of ceramide and sphingomyelin, neuron-neuron synaptic transmission, regulation of dendritic spine development, the neuronal stem cell population maintenance, brain development, motor learning, branching morphogenesis of a nerve, axonogenesis (all p < 0.05). And predicted cell components also include neuron projection, dendritic spine synapse, synaptic vesicle membrane, and postsynaptic membrane. Enrichment of the KEGG pathway showed that the target genes of miR-9-5p were enriched in the neurotrophin signaling pathway and estrogen signaling pathway. It is worth mentioning that, Taken together, the enrichment analysis and hub genes indicated that miR-9-5p and miR-4467 are likely to play a role in neurodevelopment and synaptic transmission via the regulation of target genes, which could contribute to the occurrence and development of schizophrenia. In addition, considering the involvement of target gene groups in the regulation of neurodevelopment, it further proves the signi cance of selecting rst-episode schizophrenia patients to explore the etiology and pathogenesis of schizophrenia.
For miR-328-3p and miR-144-3p whose expression changes were also detected in the sequencing samples, studies have been veri ed in autistic spectrum disorder patients (Kichukova et al. 2017;Nt et al. 2018) and Alzheimer's disease (Zhou et al. 2019). However, there is no evidence to support the abnormal expression of miR-328-3p or miR-144-3p in FES patients, and no signi cant changes in the expression of the two were found in our validation set, the trends were different from the sequencing results (p > 0.05).
In this regard, we considered that high-throughput sequencing and qRT-PCR are two different experimental platforms. High-throughput sequencing is used for large-scale screening, re ecting the overall gene expression trend of the samples, and there is a certain probability of false positives. There is no guarantee that the trend of the gene expression is exactly the same as that of qRT-PCR.
In spite of our novel ndings of alterations of miR-4467 and miR-9-5p in rst-episode schizophrenia patients. This study has some limitations that need to be acknowledged. First of all, our case-control study design has determined the fact that it cannot re ect the causal relationship between miRNA expression level and schizophrenia. Secondly, although we have enrolled enough samples based on the pre-experiment results and statistical power requirements, larger sample will de nitely help us provide more valuable data, we are still recruiting. Of further interest are the expression level of miR-9-5p and miR-4467 predicted targets in the peripheral blood of FES patients, as well as the binding sites of miRNA and target genes. Are these hub genes directly regulated by miR-9-5p and miR-4467? What are their binding fragments? These problems need to be explored through subsequent investigations at the cellular level.
In conclusion, miR-9-5p and miR-4467 in peripheral blood are found to be potential biomarkers for early diagnosis of schizophrenia, and might affect the onset and development of SCZ by target regulation of neurodevelopment-related mRNAs.

Declarations
Author contributions Declarations of interest: The authors declare no con ict of interest.  The volcano plot of differentially expressed miRNAs.  Gene ontology analysis of miR-9-5p (A) and miR-4467 (B) target genes.

Figure 4
Top 10 KEGG pathway of miR-9-5p target genes. Interaction between proteins encoded by miR-9-5p (A) and miR-4467 (B) target genes. The size and color of each edge are determined by combined-score, and the size of node is determined by MCODE-score.