Participants
We recruited patients from August 2013 to January 2017 from the inpatient units at The Menninger Clinic, Houston, TX, USA. All participants were evaluated during the admission week using the Structured Clinical Interview for DSM-IV (SCID-I/II ) [35]. We rated suicidal ideation based on the Columbia-Suicide Severity Rating Scale (C-SSRS) version 1/14/09 [36, 37] at admission and biweekly thereafter. The C-SSRS assesses suicidal ideation, plans, and actions including attempted suicides and aborted or prevented suicide attempts. We classified a C-SSRS score of one or above as having SI [37]. We evaluated sixty-eight patients for this study. These included 42 patients with SI (C-SSRS > 1) upon admission and who had no SI four to six weeks later (SI recovery group) and 26 depressed comparison patients without SI both at admission and four to six weeks later (non-SI; C-SSRS = 0). The admitting diagnoses were major depression and anxiety disorders with over half having histories of substance use disorders (SUD) (Table 1). None of the SUD patients required medical detoxification or had SUD-induced affective disorders or psychosis (SCID-I). We also assessed anxiety and depressive symptoms at admission and then biweekly using the Patient Health Questionnaire (PHQ). The PHQ includes 9-item depression and 7-item anxiety scales with items scored from 0 to 3 for maximum total scores of 27 and 21, respectively [38]. Before entering the study, all participants signed an informed consent document approved by the Institutional Review Board of Baylor College of Medicine.
TABLE 1 Demographic, Psychiatric and Medication Comparison of SI to non-SI
Rna Isolation, Library Preparation And Mirna Sequencing
We collected 10 ml of whole blood (EDTA anticoagulant) and processed it within one hour using the Gentra Puregene blood kit (Qiagen, Germantown, MD) per manufacturer protocol. Blood was spun at 2,500xg for 10 minutes. Plasma was removed and stored at -80o C until analysis. We extracted total RNA from plasma using the Qiagen miRNeasy serum/plasma kit (Qiagen, Germantown, MD). The miRNA concentration was quantified using an Agilent 2100 bioanalyzer (Santa Clara, CA), and sequencing libraries were constructed using the Qiagen Qiaseq miRNA library kit (Qiagen) according to the manufacturer’s protocol. Library quality was assessed using the Agilent 4200 tape station to ensure the miRNA-sized library was approximately 180 nt in length. The sample was loaded onto an Illumina Nextseq500 instrument (San Diego, CA) and the miRNA-sized library was sequenced.
Sequencing Data Analysis
We used the global miRNA expression profiles to create a list of expressed miRNAs. The raw sequencing reads were trimmed using Cutadapt [39] to remove adapter, primer, and poly-A sequences. Reads with fewer than 15 nt were discarded. We identified known miRNAs using miRbase v21 [40] and miRdeep2 v2.0.0.8 [41, 42] and constructed a miRNA expression matrix using the number of reads mapped to each miRNA. We determined DE miRNAs using the R package DESeq2 (v1.26.0) [43]. We considered miRNAs with DESeq2 normalized counts > 10, which occurred in at least 50% of libraries to be expressed, and we used these miRNAs in DE and for downstream analyses. We considered differences in miRNA expression between admission and SI recovery significant at Benjamini and Hochberg corrected p < 0.05 and |log2 Fold Change| > 0.25.
Real-time Quantitative Reverse Transcription Pcr (Qrt-pcr)
Significantly expressed DE miRNAs from the sequencing data analysis were validated by qRT-PCR using TaqMan Advanced miRNA Assay kit (ThermoFisher, Waltham, MA). Total miRNA was extracted from plasma utilizing the Qiagen miRNeasy serum/plasma kit (Qiagen). Complementary DNA (cDNA) was synthesized using the Reverse Transcription Taqman advanced miRNA cDNA synthesis kit (ThermoFisher). Real-time PCR used the QuantStudio Real-Time PCR system (ThermoFisher). For PCR assays, AmpliTaq Fast DNA polymerase (ThermoFisher) was activated at 95°C for 2 minutes followed by 40 cycles of 2-step amplification at 95°C for 1 s and 60°C for 20 s. The qRT-PCR reactions for each sample were run in triplicate. Raw data were collected with cycle quantification values (Cq values). Reference miRNAs were selected from the sequencing data. Stably-expressed miRNAs were determined using the GeNorm algorithm [44] with the top “stability Factors”. Two stably expressed miRNAs, has-miR-191-5p and has-miR-126-3p, were incorporated as reference normalizers. miRNA expression levels were calculated using the 2−ΔΔCq method after normalization to the mean expression levels of hsa-miR-191-5p and hsa-miR-126-3p [45].
Dual-reporter Luciferase Assay For Mirna Functional Activity
For the wild-type Clusterin 3’UTR insert, a region of 289 bps (from 1541 to 1830, 3’UTR) including the two predicted binding sites of miR-424-5p, was PCR amplified from human genomic DNA. We used a similar procedure for the predicted binding site of miR-10b-5p on the SDC1 (syndecan-1) 3’ UTR. Syndecan-1 is an important cell adhesion molecule belonging to the family of syndecans, which are transmembrane heparan sulfate proteoglycans (https://www.proteinatlas.org/ENSG00000115884-SDC1). Each were cloned into the NotI restriction site of the psiCHECK2 vector using Gibson assembly (NEB - Ipswich, MA. (Cat#E5510S); Sigma-Aldrich - Saint Louis, MO. (Cat #HMC0002-5NMOL); ThermoFisher- Carlsbad, CA. (miR-424-5p mimic (Cat#4464066, Assay ID MC10306) or miR-10b-5p mimic (Cat# 4464066, Assay ID MC11108); Promega - Madison, WI (Cat#E1910)). Colony PCR was performed to select clones containing the plasmid with the insert. The clones were verified using Sanger sequencing. A mutant plasmid was created using PCR based mutagenesis to change the seed sequence of the predicted binding sites (from tgctgct → tAAtTAA and ttgctgct → AtAAtTAA).
We plated 0.03x106 HEK 293T cells/ well in a 96-well plate with at least three technical replicates. 100 ngs of plasmid was transfected alone (NT) or co-transfected with NC microRNA (Sigma-Aldrich, Cat #HMC0002-5NMOL) or with miR-424-5p mimic (ThermoFisher, Cat#4464066, Assay ID has-miR-509-5p) at 50nM. Transfections were carried out using Lipofectamine 2000 (ThermoFisher, Cat#11668027). 48 hours post transfection, the assay was performed using the dual luciferase reporter assay kit (Promega, E1910). Data represents renilla luciferase signal normalized to the firefly luciferase signal from three biological replicates. Error bars represent SD. Statistical significance was determined by unpaired (Mann-Whitney), two-tailed, non-parametric t-tests using GraphPad Prism.
Mirna Target Gene Prediction And Functional Enrichment Analysis
We predicted miRNA target genes using open access databases. The experimentally validated miRNA-target interactions were developed by miRTarBase V8.0 using a web-based tool miRNet [46], a comprehensive knowledge base integrating high-quality miRNA-target interaction data from eleven databases. Our target gene list was uploaded to DAVID (https://david.ncifcrf.gov/summary.jsp) for functional enrichment analysis to identify the biological processes of the miRNA target genes [47, 48]. The tissue expression and the Kyoto encyclopedia of genes and genome (KEGG) pathways were utilized for annotation visualization. We selected enriched pathways based on the target genes with p values < 0.1 (default cutoff) and more than 3 target genes in that pathway.
Protein-protein Interaction Networks Construction
We chose the miRNA target gene encoded proteins for construction of the protein-protein interaction (PPI) network using STRING online software (https://string-db.org/) with a combined score > 0.4 and the interacted protein amount > 1 [49]. We then visualized the PPI network using Cytoscape V3.8.0 software [50] and identified Hub genes as the nodes with higher degrees (Degree > 40) using the CytoHubba function. We established miRNA-hub gene networks from common hub genes and miRNAs.
Clusterin Elisa Assay
Plasma clusterin protein levels were evaluated using an enzyme-linked immunosorbent assay (ELISA) kit purchased from R&D Systems (Cat. DCLU00, Minneapolis, MN) following the manufacturer’s instructions to detect clusterin concentrations (ug/ml). We defined a statistically significant difference as p < 0.05 (2-tailed paired T-test).
Statistical analysis
We performed statistical analyses using the Statistical Package for Social Sciences (SPSS) for Windows, version 23 (IBM, Armonk, New York) and RStudio version 1.2.1335 (RStudio, Boston, MA). Demographic variables in SI subjects and non-SI subjects were compared with χ2 tests for qualitative variables and T-tests for quantitative variables. miRNA expression levels as determined by qRT-PCR were evaluated using the Wilcoxon signed-rank test with the miRNA data presented as fold-change of the miRNA expression in the recovered SI sample compared to their levels at admission. Statistical significance required two-tailed p < 0.05. Significance of multiple comparisons was estimated using the false discovery
rate (FDR).