Human Study
Participants
Adolescents, ages 12-19, were recruited in the greater New York City area. Participants under age 18 provided written assent, and a parent or guardian gave written informed consent; participants 18 years and older provided written informed consent. The study was approved by the Institutional Review Board of Icahn School of Medicine at Mount Sinai.
Inclusion and Exclusion Criteria
Inclusion criteria for adolescents with psychiatric symptoms: presence of one or more mood or anxiety symptoms based on diagnostic evaluation. Symptoms were clinically significant but were not required to meet the severity threshold for a full diagnosis.
Inclusion criteria for healthy control adolescents: no past or current psychiatric conditions or clinically significant symptoms.
Exclusion criteria for all participants: 1) any physical or neurological conditions; 2) estimated IQ<80; 3) a positive drug toxicology test; 4) a positive pregnancy test; 5) current psychosis, pervasive developmental disorder, or substance abuse; 6) psychotropic medication use in last 1-3 months at baseline visit, depending on drug half-life; 7) any recent inflammatory illnesses, including the common cold; and 8) any recent anti-inflammatory medication use, including over-the-counter remedies.
Clinical Assessments
All participants received diagnostic evaluations using the Schedule for Affective Disorders and Schizophrenia for School-Age Children–Present and Lifetime Version (KSADS-PL) [36], administered by a trained child and adolescent psychiatrist or a clinical psychologist. Additional assessments included the Children’s Depression Rating Scale-Revised (CDRS-R) [37] to measure the presence and severity of depression symptoms and the Kaufman Brief Intelligence Test (K-BIT) [38] to estimate overall IQ. Clinical symptom severity was quantified using the self-rated Beck Depression Inventory (BDI) [39] and Multidimensional Anxiety Scale for Children (MASC) [40].
Trier Social Stress Test Procedures
Participants arrived in the early morning after fasting ≥12 hours. A small cannula was inserted into the vein to allow for two blood draws. The first was conducted before the TSST and the second was conducted within 5-10 minutes after TSST completion. Participants completed a slightly modified version of the TSST, designed for children and adolescents [41], composed of two sections: a brief preparation period followed by a mock interview in which the participant told a story and performed mental arithmetic before a panel of three adult judges. The youth TSST protocol thus maintains important elements of a stress-inducing task, including a threat to the social self, uncontrollability, and unpredictability. This protocol has been shown to raise blood pressure, heart rate, and salivary cortisol in children and adolescents, even when controlling for the novelty of visiting a laboratory [42], and has been shown to reliably induce stress in adolescent populations in clinical settings [43]. Following the TSST, participants rated how anxious they felt during the task from 1 (“Not at all”) to 6 (“Extremely”) using a visual analog scale (VAS).
Immune Biomolecule Assessment
Procedures follow those described in our previous immunological studies [7,8]. Whole-blood samples were cultured for six hours under two conditions: using standard culture medium alone and using standard culture medium plus LPS (0.1µg/ml). Supernatants were then harvested and analyzed for immune biomarkers using a Luminex-200 system and the xMap Platform (Luminex Corporation, Austin TX) following manufacturer recommendations. Levels were determined in duplicate 25μL volumes of supernatant using multiplex panels (Multiplex High Sensitivity 41-plex Human Cytokine/Chemokine Panel, Millipore Corp.) and reported as median fluorescence index (MFI) values; analyses used the mean of the two duplicate MFI measurements per analyte. Lower and upper detection limits for assays were 3.0pg/mL and 15ng/mL, respectively. Multiplex assays targeted 41 unique immune biomolecules (see Table 2), including the cytokines IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, interferon (IFN)-γ, and TNF-α.
Statistical Analyses
Immune activation was evaluated using Matlab 2017a. Lilliefors goodness-of-fit tests indicated that levels of most biomarker were not normally distributed in our sample. Data were therefore normalized using the Box-Cox method to allow the use of parametric analyses. The effects of TSST, LPS, and TSST×LPS interaction were then tested using repeated-measures analysis of variance (rANOVA) in a within-subject 2×2 design, with diagnostic status (clinical or healthy control participant), age, sex, and body mass index (BMI) included as covariates. To further evaluate potential associations between immune biomarkers and TSST-induced stress, rANOVA analyses were repeated including participants’ self-reported anxiety during the TSST (1-6 VAS scores) as an additional covariate. Results were controlled for multiple comparisons using Bonferroni correction (two-tailed pBonferroni≤0.05/41≈1.2×10-3). Complete subject-level data used for rANOVA, including covariates and analyte levels at each timepoint/condition, are provided in Supplementary Tables S1-S4. Post hoc sensitivity analyses conducted in G*Power v3.1.3 indicated 96% power to detect moderate effects (Cohen’s f=0.25) and 80% power to detect effects as small as f=0.20 in our sample.
Animal Study
Mice
All experimental protocols were approved by the Institutional Animal Care and Use Committee of Albert Einstein College of Medicine. Seven-week-old male experimental mice (C57BL/6N) were purchased from Envigo. CD-1 male retired breeders over 16 weeks of age were purchased from Charles River. After arriving to our colony, all animals were single housed with ad libitum access to food and water on a 12hr light/dark cycle (7AM-7PM). The experiment was performed after one week of acclimatization in our satellite facility. CD-1 mice were kept in separate room from the experimental mice to avoid habituation to their odors. Prior to the experiment, CD-1 mice were screened for aggressive behavior to identify viable resident aggressors.
The social stress paradigm was performed when experimental mice were 60 days (eight weeks) old, which corresponds to an approximate human age of 16 years (see Ch. 20 in [44]) and may be considered late adolescence or early adulthood, depending on the source. This age was chosen based on previous findings that acute social stress paradigms reliably induce behavioral changes in mice aged 50-61 days but not in younger mice aged 29-40 days [45,46]. Similarly, LPS exposure increases neural cFos expression and serum corticosterone levels in young adult mice aged 70 days but not in peripubertal mice aged 42 days [47]. Experimental mice were limited to males to ensure the validity of the social defeat paradigm, which does not readily generalize to females [48].
Mice were randomly assigned to one of three main experimental groups, detailed below: non-stressed (NS) mice receiving injections of saline (NS-SAL) or LPS (NS-LPS) and mice exposed to the reminder social defeat (SD) paradigm receiving saline injections (SD-SAL). Timecourse data were collected at 0hr, 1hr, 6hr, and 24hr post-injection. The 6hr timepoint also included a fourth experimental group of mice exposed to the reminder SD paradigm and given LPS injections (SD-LPS). Six mice were assigned to each group at each timepoint (N=78 total).
An overview of experimental procedures in the animal study is provided in Figure 1A.
[Figure 1 here]
Reminder Social Defeat Paradigm
The social stress paradigm was an acute version of the reminder SD protocol [49]. Each experimental C57BL/6N mouse in the SD branch was introduced to the cage of a resident CD-1 aggressor for five minutes, during which the intruders were attacked and exhibited avoidance, submissive, and defensive behaviors (SD exposure). After 48hr, each experimental mouse was returned to the same resident aggressor’s cage for 5min but placed behind a partition, enabling the experimental mouse to smell and see the dominant CD-1 mouse but preventing physical contact between the animals (“reminder” SD exposure). Mice in the non-stressed (NS) branch were not exposed to resident CD-1 aggressors.
Behavioral Validity of Reminder Social Defeat Paradigm
Behavioral validity data for the reminder SD paradigm are included from a separate cohort of 20 male C57BL/6N mice, illustrated in Figure 1B. Mice were introduced to the cage of a resident CD-1 aggressor for 5min (SD exposure). After 24hr, half of mice were returned to the same resident aggressor’s cage behind a partition for 5min (reminder SD exposure) and half received no reminder SD exposure. After another 6hr, behavioral assays were performed on all mice. Habituation was assessed by returning mice to the same resident aggressor’s cage alone for 5min and measuring the fraction of time spent on each side. Sociability was then assessed for 5min by placing a conspecific mouse on one side of the cage and a toy mouse on other side of the cage and measuring the fraction of time spent on each side. Locations of conspecific and toy mice were counterbalanced.
Induction of Systemic Inflammation
Following completion of the reminder SD protocol, mice in the LPS branch received a single intraperitoneal (i.p.) injection with LPS (Escherichia coli 005: B5, No. L-2880 Sigma-Aldrich, St. Louis MO) at a dose of 0.83mg/kg dissolved in 200μl of sterile, pyrogen-free physiological saline. LPS dose was selected to induce acute systemic inflammation based on earlier findings [50,51]. Mice in the SAL branch received an equivalent i.p. injection of saline alone as a negative control. Experimental mice were euthanized 0hr, 1hr, 6hr, and 24hr after injection in the NS-SAL, NS-LPS, and SD-SAL groups and 6hr after injection in the SD-LPS group.
Immunohistochemistry of cFos
Mice were anesthetized with an i.p. injection 1ml Avertin (480mg/kg of tribromoethanol/ isopropyl alcohol) and transcardially perfused with ice-cold 4% paraformaldehyde in phosphate buffer (pH=7.4). Brains were removed and post-fixed for 24hr in the same fixative and then immersed in 20% and 30% sucrose in phosphate buffer for 24hr each. Brains were frozen for 1hr at -80°C in Tissue-Plus™ O.C.T. Compound (Thermo Fisher Scientific, Waltham MA) and 50μm sections were cut. For antigen retrieval, sections were incubated in citrate buffer (pH=6.0) for 30min at 70oC, washed 3× in phosphate-buffered saline (PBS), and blocked in PBS containing 0.3% Triton X-100 and 5% normal goat serum. Sections were incubated overnight at 4ºC in cFos antibody 1:2000 (Abcam, ab208942). After incubation, sections were washed in PBS containing 0.2% Triton X-100 and 1% normal goat serum, then incubated for 2hr at room temperature in Alexa Fluor® 488 AffiniPure donkey anti-rabbit IgG (H+L) 1:500 and 594 AffiniPure donkey anti-mouse IgG (H+L) 1:400 (Jackson Immuno Research, West Grove PA, 711-545-152 and 715-585-150). Nuclei were counterstained with Hoechst 33342 1:3000 (Thermo Fisher Scientific). Prepared sections were slide mounted and cover-slipped with Vectashield (Vector Laboratories, Burlingame CA). Sections were observed on a Leica fluorescent microscope and Olympus Fluoview confocal microscope using 10× and 60× objectives.
As illustrated in Figure 1C, images were captured at the level of paraventricular hypothalamic nucleus (PVN), paraventricular thalamic nucleus (PVT), and basolateral amygdala (BLA). These regions were selected due to their involvement in stress response [52,53] [54,55] as well as published [54] and in-house data showing that these areas exhibit sustained upregulation of cFos following acute immune challenge. Quantification was performed using ImageJ, as described previously [55]. For cFos, cell counts were performed using a constant threshold (diameter >5μm, intensity >0.47a.u.). All quantification performed blind to experimental group using two sections per mouse, unless prohibited by lesion.
Statistical Analyses
Mouse immunohistochemical data were analyzed using Graphpad. All samples were normally distributed, as determined with a one-sample Kolmogorov-Smirnov test. Homogeneity of variance was confirmed with Levene’s test for equality of variance. The results were analyzed using two-way ANOVAs with group and timepoint as factors. For the 6hr timepoint, additional two-way ANOVAs were run with stressor (NS, SD) and injection type (SAL, LPS) as factors. Significant F values were followed by post hoc pairwise t-tests. All tests were run separately for each region and Bonferroni adjusted for multiple comparisons (two-tailed pBonferroni≤0.05/3≈0.017). Power considerations for the animal study were informed by our recent work [56-58] [58-60] as well as pilot experiments. Minimum power to detect anticipated effects was set at 80% (1-β>0.80; two-tailed α≤0.05). Sensitivity analyses performed using G*Power v3.1.3 indicated power to detect moderate-to-large group (f=0.37) and timepoint (f=0.40) effects in our sample.