Participants
Participants were Caucasian males recruited via flyers posted on the Stony Brook University campus and through advertisements in local newspapers in surrounding communities on Long Island, NY, USA. Study exclusion criteria included: age under 18 years, significant current or prior presence of psychiatric illness, smoking, presence of substance or alcohol abuse in the past six months, hormonal, mood-altering, heart disease and/or cardiovascular conditions related, or psychoactive drug medication, BMI ≥ 30. Participants were pre-screened for exclusion criteria by phone. Fifty-five participants were part of the study, which was approved by the Institutional Review Board (IRB) and Committee on Research Involving Human Subjects (CORIHS) of Stony Brook University, Stony Brook, USA. We selected 12 participants who showed a robust cortisol response to the TSST (“Cortisol Responders”; based on a criterion suggested by Miller et al. [10]) as a discovery cohort to generate repeated-measures, within-subject genome-wide mRNA and miRNA expression data.
Psychosocial Stress Procedure: TSST
All participants were scheduled for a four-hour weekday laboratory session conducted at the General Clinical Research Center (GCRC) at the Stony Brook University Hospital from 1400h to 1800h. Upon arrival, participants were informed about the study, completed the consent procedure, and underwent insertion of an intravenous catheter into the arm for the first blood draw. Participants then rested for a period of at least 45 minutes before they were introduced to the psychosocial stress paradigm. The Trier Social Stress Test (TSST) is a standardized social stress protocol that was used to induce psychosocial stress and that has been found to elicit a strong and reliable physiological response to laboratory stress [8]. Briefly, the standardized laboratory stressor consists of a 5-min preparation period, a 5-min free speech task during which participants give a mock job interview, and a 5-min mental arithmetic task wherein participants perform serial subtraction by thirteen all in front of a non-responsive audience.
Biological Measures: Cortisol
To assess cortisol levels, ten salivary samples were obtained from each participant throughout the duration of the study. Samples were collected using Salivettes® (Sarstedt) at arrival, 2 minutes before beginning the TSST, and again at 2, 10, 20, 30, 45, 60, 90- and 105-minutes post stress task. Participants were instructed to gently chew on the cotton swab for approximately one minute during each collection. Cotton swabs were then transferred to plastic containers and immediately stored at -20oC until shipment to the analyzing laboratory (N. Rohleder, Brandeis University, Waltham, MA). Salivary cortisol concentrations were measured using a commercially available chemiluminescence immunoassay (CLIA; RE62019; IBL International, Toronto, Canada). Inter-assay variability was 3.3%, and intra-assay variability was 3.4%. Salivary cortisol concentrations have been shown to highly correlate with serum cortisol concentrations, making salivary cortisol a reliable indicator of HPA activity [11]. Participants had no oral intake except for water for 3 hours before beginning the study session. Cortisol response to the stress paradigm was assessed by subtracting the highest mean cortisol concentration following the stressor minus the baseline cortisol concentration (two minutes before the TSST). Using a criterion suggested by Miller et al. [10], those participants who showed a cortisol increase of at least 1.5 nmol/l from baseline were assigned in the “Cortisol Responder” group (n=33, mean cortisol increase of 8.55 nmol/l (±5.63), while those who did not show an increase of at least 1.5 nmol/l from baseline were placed in the “Cortisol Non-Responder” group (N=22, mean cortisol increase of -3.889 nmol/l (± 6.45. The two groups differed in age (responders: 34.58 years ± 17.09; non-responders: 26.18 years ±13.44; t-test p<0.05). There were no significant differences in psychological variables of interest, as the BAI, BDI, TiCS, STAI-S and CTQ sum scores were not significantly different between the two groups (all p>.05). There was no significant difference in the number of stressful life events (SLEs) between the two groups (p > .05). Subsequent gene expression analyses with microarray profiling were conducted on the first 12 Cortisol Responders.
Biological Measures: Blood Collection
Directly upon arrival and following the informed consent procedure, an intravenous (IV) catheter was inserted into each participant’s arm for the first blood draw (Time Point A). Participants then rested for a period of at least 45 minutes before they were introduced to the psychosocial stress paradigm. The second blood draw occurred 45 minutes after the stress paradigm (Time Point B) and the third blood draw occurred 105 minutes post stress (Time Point C). Ten ml of EDTA blood (VacuetteR Greiner Bio-one) was collected at each of the three time points. Blood was immediately processed for PBMC extraction according to the Leucosep® Instruction Manual (Greiner Bio-one), and cell pellets were frozen and stored at -80oC.
RNA Isolation, Quantification, Quality Control
Total RNA was extracted from PBMCs at each time point using the Qiagen All-Prep kit with company-recommended modification to extract small RNA (Qiagen). RNA quantity was assessed using a Nanodrop Technologies ND-1000 instrument (NanoDrop Technologies). Total RNA quality was measured using the Agilent RNA 6000 Pico kit and the Agilent Bioanalyzer 2100A (Agilent Technologies), while miRNA content was resolved using the Small RNA Kit (Agilent). RNA samples were stored at -80°C.
MiRNA Microarray Profiling
mRNA microarray profiling was performed at the ShanghaiBio Corporation using Affymetrix Human Gene ST 1.0 Arrays according to the manufacturer's protocol. Briefly, 500 ng of total RNA was used in complementary DNA synthesis and labeling, followed by hybridization, washing, scanning, and image processing of GeneChips according to standard Affymetrix protocols. Quantitative image files were loaded into the Bioconductor Affymetrix linear modeling Graphical User Interface software [12] and background adjustment, quantile normalization, probe summarization, and log2 transformation was performed using the Robust Multi-Array Average method [13].
Statistical Analysis of mRNA and miRNA Microarray Data
Time-course microarray data analysis was conducted using the Linear Models for Microarray Data (limma) Bioconductor package [14]. Linear models were fitted to distinguish differentially expressed mRNAs between the three time points (A, B, C) on the Affymetrix platform. mRNA showing limma p<0.01 for at least one of the time point comparisons (B to A, C to A, B to C) correcting for multiple comparisons using false discovery rate (FDR) correction, as well as with a fold change (FC) |FC|≥ 1.3 were highlighted as genes that show greater variation in expression in response to the TSST. The fold change threshold of |FC| ≥ 1.3 was chosen as literature has conventionally utilized fold changes ≥ 1.3-2 to suggest differential expression in mRNA microarray data [15]Also, fold change of gene expression in blood has usually been less than two-fold [3, 16] and multiple studies have used a cut-off of FC > 1.2 for identifying changes of gene expression specifically in blood [16-18]. miRNAs generally show more subtle changes in expression than mRNAs, but even small expression changes can have significant biological effect as each miRNA has multiple gene targets and can influence multiple genes in a given pathway. Therefore, miRNAs showing a |FC|≥ 1.2 and limma p value<0.01 for at least one of the time point comparisons (B to A, C to A, B to C) were highlighted as showing change in expression in response to the TSST.
MiRNA/mRNA Integration and Functional Profiling
Integration of mRNA and miRNA expression was performed by the computational gene network prediction tool Ingenuity Pathway Analysis (IPA, Ingenuity® Systems). Our study showed a limited number of mRNAs that were significant according to limma p<0.01 and |FC|≥ 1.3, possibly in part due to the small study sample size and conservative statistical inclusion criteria. It is worthwhile noting that a single miRNA can subtly regulate multiple genes in a given pathway or process, having an additive biological effect. Therefore, mRNAs that showed significant changes in expression at any time contrast according to limma p<0.01 were input into IPA. In parallel, human orthologues of differentially expressed miRNAs (limma p<0.01 and |FC|≥ 1.2) were input into the IPA program as well. To integrate the two datasets, identification of negatively correlated miRNA/mRNA pairs was performed using the Target Filter Analysis in IPA. The algorithm uses experimentally validated interactions as well as predicted microRNA-mRNA interactions from TargetScan. Fold change at each miRNA and mRNA was integrated at Time B vs Time A, Time C vs Time A, and Time C vs Time B. The expression pairing analysis was used to identify negatively correlated miRNA/mRNA pairs that had experimentally demonstrated targeting or a moderate to high degree of likelihood of a targeting relationship based on prediction algorithms.