Repeated stress exposure in mid-adolescence attenuates behavioral, noradrenergic, and epigenetic effects of trauma-like stress in early adult male rats

Stress in adolescence can regulate vulnerability to traumatic stress in adulthood through region-specific epigenetic activity and catecholamine levels. We hypothesized that stress in adolescence would increase adult trauma vulnerability by impairing extinction-retention, a deficit in PTSD, by (1) altering class IIa histone deacetylases (HDACs), which integrate effects of stress on gene expression, and (2) enhancing norepinephrine in brain regions regulating cognitive effects of trauma. We investigated the effects of adolescent-stress on adult vulnerability to severe stress using the single-prolonged stress (SPS) model in male rats. Rats were exposed to either (1) adolescent-stress (33–35 postnatal days) then SPS (58–60 postnatal days; n = 14), or (2) no adolescent-stress and SPS (58–60 postnatal days; n = 14), or (3) unstressed conditions (n = 8). We then measured extinction-retention, norepinephrine, HDAC4, and HDAC5. As expected, SPS exposure induced an extinction–retention deficit. Adolescent-stress prior to SPS eliminated this deficit, suggesting adolescent-stress conferred resiliency to adult severe stress. Adolescent-stress also conferred region-specific resilience to norepinephrine changes. HDAC4 and HDAC5 were down-regulated following SPS, and these changes were also modulated by adolescent-stress. Regulation of HDAC levels was consistent with the pattern of cognitive effects of SPS; only animals exposed to SPS without adolescent-stress exhibited reduced HDAC4 and HDAC5 in the prelimbic cortex, hippocampus, and striatum. Thus, HDAC regulation caused by severe stress in adulthood interacts with stress history such that seemingly conflicting reports describing effects of adolescent stress on adult PTSD vulnerability may stem in part from dynamic HDAC changes following trauma that are shaped by adolescent stress history.

www.nature.com/scientificreports/ within blocks), and treatment groups were evenly distributed during the first and last hours of the testing. To minimize disturbance, the experimenter was not in the room during fear conditioning procedures or adolescent stressor administration. Procedures were approved by the VA Ann Arbor Healthcare System Institutional Animal Care and Use Committee (#1312-004); all procedures were performed in accordance with the relevant guidelines and regulations.
Repeated variable stress in adolescence. Rats were exposed to repeated variable stress in mid-adolescence, from 33 to 35 postnatal days of age. The repeated variable stress paradigm encompassed three stressors, each representing a different modality of predation cues (visual: a swooping hawk model, olfactory: fox urine, auditory: large cat vocalizations, stressors described in Table 1). Rats encountered 1 stressor per day, for 3 consecutive days. Stressor timing was varied (i.e. unpredictable) and counterbalanced such that half of the rats from each age group experienced 2 stressors during the light phase and 1 during the dark phase with the other half experiencing 1 stressor during the light phase and 2 during the dark phase.
Single prolonged stress. SPS has been used for two decades to model PTSD traits 63,64 ; reviewed in 50 .
In SPS, rats are exposed to three stressors in succession followed by isolation for 7-days [additional detail in supplementary methods and 44,64,65 ]. Briefly, rats were restrained for 2 h, then forced to swim (23-24 °C) in a 68 × 56 × 45 cm opaque plastic container in groups of 6-8 for 20 min. Rats were then towel-dried and given 15 min to recuperate with a heat source. Next, rats were exposed to diethyl ether vapors in a desiccator until loss of consciousness. Finally, rats were individually-housed in clean cages and undisturbed for 7-days 44,67 . SPSinduced neuroendocrine effects, including HPA negative feedback and glucocorticoid receptor mRNA expression, are only evident a 7-day quiescent period of SPS 63,67 . Isolation can have neuroendocrine effects, thus control rats were isolated to account for potential housing effects which could otherwise confound group comparisons 68 .
Fear learning, extinction, and extinction retention testing. SPS can induce an extinction retention deficit, in which fear is enhanced after a fear association has been extinguished, indicating the dominance of fear memory over extinction memory 44,65,67 . To test extinction retention, rats were trained on the 8th day following SPS to associate a tone with a shock across five shock-tone pairings in a fear conditioning chamber (day 1: fear conditioning; shock features: 1 s, 1 mA; tone: 10 s, 1 kHz, 80 dB, 60 s inter-trial-interval 65,69 ). Freezing responses were quantified as a proxy of fear 44,70 . On the 9th day after SPS, rats were presented with the same tone 30 times without the shock to extinguish fear responses in a second context (day 2: fear extinction). The second context was differentiated by novel auditory, visual, tactile, and olfactory cues 65,69 . Finally, rats were returned to the extinction context on the 10th day after SPS and were re-exposed to the tone without the shock for 10 presentations (day 3: extinction retention). Additional details are provided in the supplementary methods, based on 65 .

Figure 1.
Timeline of procedures. Repeated variable stress consisted of repeated, unpredictable exposure to visual, olfactory, and auditory predation cues (a swooping hawk model, fox urine, large cat vocalizations). The single prolonged stress (SPS) model is defined by three stressors in succession (restraint, forced swim, ether exposure). Following SPS, rats were socially isolated for seven days. During fear learning testing, rats were first trained to associate a tone with a shock, then the tone was repeatedly presented in a novel context to facilitate extinction learning. Rats were then returned to the second context to test the retention of extinction learning. The day following extinction retention testing, brains were collected for region-specific measurement of HDAC4, HDAC5, NE, and NPAS4. All groups were age-matched for SPS, fear learning, and time in the laboratory; groups 2 and 3 were reared in the laboratory under control conditions. www.nature.com/scientificreports/ Freezing was defined as immobility, > 1 s, with or without small pendulum-like head movements with feet/body/ neck immobility without vibrissae movement 69 , as this is indicative of fear in rats 71 and other small mammals 72,73 . Brain dissection for neurochemical analysis. Brains were harvested after rapid decapitation without anesthesia and immediately frozen on dry ice for later processing, on the day following extinction retention testing [methods based on 69 ]. To obtain tissue punches of specific regions, brains were thawed at − 20 °C for 10 min. Brains were sliced into 2 mm coronal sections using a chilled stainless-steel rat brain matrix. Bilateral tissue punches (1.5 mm) were obtained from the PLC, dorsolateral striatum, amygdala, and dorsal hippocampus, in accordance with the Paxinos and Watson Rat Brain Atlas, and frozen at − 80 °C for subsequent analysis. Punch hemisphere was randomized at the individual-level for high-pressure liquid chromatography (HPLC) and alternative punches were used for immunoblotting.
Norepinephrine analysis with high pressure liquid chromatography (HPLC). Whole tissue NE levels were measured with HPLC [methods based on 69,74 ]. Tissue punches were suspended in 50 μL of 0.2 N HClO4, then sonically disrupted and centrifuged (4 °C; 12,300 rpm; 10 min). Next, 25 μL aliquots of supernatant were loaded into a Dionex Ultimate 3000 HPLC system for analysis (Thermoscientific, Waltham, MA). Thermoscientific TEST Mobile Phase flowed in the column at a rate of 0.6 mL/min; containing acetonitrile, phosphate buffer, and an ion-pairing reagent. Coulometric electrochemical detection was achieved with a dual electrode cell set at − 175 mV (reference) and 300 mV (working). Chromatograms were analyzed using Dionex Chromeleon software (version 7); a detection threshold was set at 3 times the average height of 4 solvent peaks (neurochemicals below this threshold were omitted from further analysis). Absolute values of NE were determined by comparison with external standard (Sigma-Aldrich, St. Louis, MO) run in parallel and in duplicate at the beginning and the end of each run. NE levels were corrected using frozen tissue weight to obtain total concentration, expressed as ng neurochemical/mg tissue weight.
Data analysis. Percent freezing during each fear learning testing phase was analyzed with a separate repeated measure analysis of variance (RMANOVA), with stress condition and time as fixed effects. Fear extinction (30 trials) and extinction retention (10 trials), due to their length, were separated into an early phase (first half of trials) and late phase (second half of trials) 44,69,79 . If a main effect was detected, groupwise univariate ANOVAs were used to compare each group directly. One rat from the SPS-only group was omitted from behavioral analysis because the video system malfunctioned and did not provide a recording for that animal. To evaluate NE, HDAC4/5, and NPAS4, univariate general linear models were used with adolescent-stress and SPS as fixed effects. If analytes were below the threshold for detection in HPLC the individual measurement was omitted from analysis, these included: one SPS-only animal for the hippocampus; one control and one SPS-only animal in the prelimbic cortex; three control and two rats from each SPS group in the striatum. For the NPAS4 PLC nitrocellulose membrane, two samples were obscured by hyper-staining and were omitted: one SPS-only and one adolescent-stress animal. Analyses were run using IBM SPSS Statistics v.24; values are reported as means ± standard error.

Results
Fear learning. All groups showed equivalent increases in fear during the fear conditioning phase (treat- In both phases animals exposed to AS and SPS did not differ from controls (phase 1: AS + SPS vs. control: In the first half of the extinction retention testing there was a main effect across the three treatments (F 1,29 = 3.32, P = 0.05, interaction and time effects P > 0.05; Fig. 2C). Subsequent group comparisons showed that SPS enhanced freezing during the first half of extinction retention compared with unstressed rats (SPS vs. control: F 1,16 = 6.94, P = 0.02, interaction: F 1,16 = 2.42, P < 0.06), but rats exposed to AS and SPS did not differ from either the SPS only or unstressed groups (P > 0.05). For the second phase of extinction retention testing there was a main effect of treatment (F 1,29 = 3.70, P = 0.04, no interaction or time effects, P > 0.05). SPS rats continued to show elevated freezing responses compared with unstressed rats (SPS vs. control: F 1,16 = 5.46, P = 0.03, no interaction detected). However, the combination of AS and SPS decreased freezing compared with SPS alone, i.e. increased extinction retention (SPS vs. AS + SPS: F 1,22 = 4.09, P = 0.05, no interaction detected). Further, the extinction retention freezing scores of rats exposed to AS prior to SPS did not differ from unstressed rats (AS + SPS vs. control: F 1,20 = 0.50, P = 0.49, no interaction detected).

Norepinephrine levels.
In the PLC, SPS exposure alone did not affect NE levels (SPS main effect: F 1,29 = 2.46, P = 0.13). However, PLC NE was elevated by the combination of AS and SPS (AS main effect: F 1,29 = 4.78, P = 0.04, Figure 2. Effects of repeated variable stress during mid-adolescence on fear cognition after traumatic-stress in adulthood. (A) All rats showed equivalent fear learning regardless of stress history. (B) Fear behavior during fear extinction learning was heightened in rats exposed to SPS stress in adulthood compared with control rats ( # P < 0.05). Exposure to the combination of adolescent-stress and adult SPS increased freezing in the early phase of fear extinction learning, compared with control rats ( + P < 0.05), but this effect abated over time and was not present during the second half of extinction learning. (C) Exposure to adult SPS without prior adolescent-stress enhanced fear behavior during both extinction retention test phases ( # P < 0.05). Conversely, the combination of repeated variable stress in adolescence and adult SPS enhanced the retention of safety information during the late phase of extinction retention compared with traumatic stress alone (^P < 0.05).

Figure 3.
Norepinephrine levels in brain regions mediating fear cognition in adult male rats that were exposed to either repeated variable stress in adolescence followed by SPS, only age-matched SPS exposure, or unstressed control conditions. (a) indicates an effect of adolescent-stress exposure; (b) indicates an effect of SPS exposure in adulthood; (*P < 0.05, + P = 0.11). In the prelimbic cortex of young adult rats, following extinction retention testing, norepinephrine levels were elevated by the combination of repeated variable stress and adult SPS but were not affected by adult SPS alone. In the striatum, adult SPS elevated norepinephrine, but this effect was reversed by prior exposure to repeated variable stress in mid-adolescence. In the hippocampus, exposure to adult SPS decreased levels of norepinephrine; there was a trend towards a norepinephrine increase from adolescent-stress exposure.  Fig. 4). Compared to controls, SPS decreased PLC HDAC4 levels by 34%, whereas the combination AS + SPS increased PLC HDAC4 levels by 4%. Similarly, compared with the control condition, PLC HDAC5 levels were decreased 49% by SPS and 26% by AS + SPS, but no AS effect was detected (SPS main effect: F 1,23 = 6.08, P = 0.02; AS main effect: F 1,23 = 1.53, P = 0.23). The HDAC5 target NPAS4 was not affected by either stress manipulation in the PLC (P > 0.51, Supplementary Fig. S1).
In the amygdala, neither AS nor SPS had detectable effects on HDAC4/5 levels (P > 0.15). Thus, NPAS4 was not evaluated in the amygdala.

Discussion
We predicted that adolescent-stress exposure would exacerbate extinction retention deficits as well as epigenetic and noradrenergic changes seen in animals exposed to trauma-like stress in adulthood. Contrary to our prediction, adolescent-stress buffered the adverse effects of adult severe stress on extinction retention and HDAC levels ( Fig. 2A,B). In line with our predictions, adolescent-stress increased PLC NE following adult SPS, suggesting that adverse adolescent conditions lastingly regulate noradrenergic responses to severe stress in adulthood. In the other regions tested, NE did not conform to our prediction: SPS elevated striatal NE but this effect was not seen if SPS was preceded by adolescent-stress, suggesting buffering effects of adolescent-stress. In the hippocampus, NE was decreased by adult stress exposure, without a detectable effect of adolescent-stress. These results support dynamic, region-specific effects of adolescent-stress on adult noradrenergic signaling rather than additive effects of the combined stress manipulations (Fig. 3). Similarly, adolescent-stress prior to adult-stress had systematic buffering effects on HDAC4 and HDAC5 levels in brain regions that mediate cognitive effects of traumatic stress (Fig. 4). Thus, exposure to stress in adolescence has the capacity to shape the cognitive, noradrenergic, and epigenetic effects of future stress in adulthood.
The effects of adolescent-stress and adult SPS on NE levels were region specific but were not additive in any region tested. In the PLC, the combination of adolescent-stress and adult SPS exposure increased NE, which is consistent with prior evidence that stress exposure increases PLC NE 80,81 , yet SPS alone did not affect PLC NE [similar to 61 ]. Thus, regulation of stress-induced PLC NE could be a mechanism by which adolescent-stress precipitates the emergence of adverse cognitive effects of traumatic stress. Our findings extend prior evidence that adolescence may be a period of increased stress sensitivity in NE regulation compared with adulthood 82 , and support that adolescence is a sensitive period for shaping adult responses to stress through region-specific NE regulation. In the hippocampus, SPS decreased NE; hippocampal NE modulates synaptic plasticity and is necessary for the retrieval of contextual memories such that decreased hippocampal NE could inhibit extinction retention 83,84 . In the striatum, adolescent-stress appeared to buffer effects of SPS on NE levels, concurring with the group patterns of extinction retention performance. Post hoc analysis did not reveal individual-level relationships; a design powered for individual-level analysis could elucidate these patterns. The functional significance of increased striatal NE is unclear, but it should be noted that NE changes are dynamic and exposure to shocks, such as those administered during fear conditioning, triggers release of NE from the locus coeruleus thereby increasing NE 58,84 . However, a time-course study has yielded insights into the timing of NE release following shock and has shown a short-term depletion in NE across the brain (including in the PFC and hippocampus), except in the striatum 85 . Overall, changes in NE are dynamic and mediated by pre-existing NE levels and the specifics of task stimuli 86 . To elucidate the relationship between effects of stress history on HDAC4/5 levels and changes in NE demonstrated by the current results, future studies could leverage pharmacological manipulation of NE or HDAC4/5 levels and time-course measurements in limbic and frontal regions and the locus coeruleus. Defining the capacity for stress history to shape effects of trauma on NE is essential given that (1) NE regulation has been implicated in various features of PTSD, including aberrant fear extinction 48 , and (2) therapeutic drugs to either elevate or reduce NE transmission have both had varied success in off-label treatment of PTSD, highlighting key gaps in current knowledge 86 . Of interest, administration of a systemic β-noradrenergic receptor antagonist after fear conditioning in male rats can reduce fear behavior during extinction learning and enhance extinction retention, demonstrating a role of NE in extinction learning and memory 87 .
Consistent with our predictions and prior clinical evidence of decreased HDAC4 in PTSD 33 , our results demonstrate that SPS decreased HDAC4/5 in brain regions mediating cognitive effects of traumatic stress. Contrary to our prediction, we found that exposure to adolescent-stress prior to adult SPS buffered effects on www.nature.com/scientificreports/ Figure 4. Histone deacetylase (HDAC) 4 and 5 levels in brain regions mediating fear cognition in adult male rats that were exposed to either repeated variable stress in adolescence followed by SPS in adulthood, only agematched SPS exposure, or unstressed control conditions. (a) indicates an effect of adolescent-stress exposure; (b) indicates an effect of adult SPS exposure; (*P < 0.05, + P = 0.10). In the prelimbic cortex, SPS exposure decreased levels of HDAC4 and HDAC5, while prior exposure to repeated adolescent-stress mitigated these effects (HDAC4 levels were shaped by exposure to adolescent-stress; HDAC5 levels were decreased by SPS).
In the hippocampus, HDAC4 and HDAC5 levels were decreased by SPS, but this effect was reversed by prior adolescent-stress. In the striatum, exposure to repeated variable stress in mid-adolescence followed by SPS increased HDAC4 and HDAC5, whereas SPS alone decreased HDAC4 and HDAC5; however, only adolescentstress had a statistically detectable effect. In the amygdala, neither mid-adolescent stress nor SPS had detectable effects on HDAC levels (P > 0.15). www.nature.com/scientificreports/ HDAC4/5. This is in contrast to apparent cumulative effects of adverse childhood experiences on methylation 9-12 , and reflects evidence that while DNA methylation is thought to lead to stable gene repression, certain histone modifications are reversible and regulate dynamic pathways that may present novel therapeutic targets 88 . Our current results suggest that HDAC4/5 expression can be dynamically modulated by both current stress and stress history. Adolescent-stress could buffer SPS-induced changes in HDAC4/5 through three possible mechanisms: (1) preventing changes in HDAC4/5 following SPS, (2) reversing effects of SPS after a temporal delay, or (3) opposing effects that precede SPS exposure (i.e. lasting HDAC4/5 increases). The latter possibility is less congruent with accumulating evidence of dynamic HDAC regulation in response to environmental conditions 31,33 . Adolescentstress could prevent (1) or reverse (2) effects of SPS on HDAC4/5 by lastingly modulating the sensitivity by which HDAC4/5 are synthesized or degraded. HDAC4 degradation is regulated by the ubiquitin-proteasome system, with stress-dependent activity enabling cells to withstand stress 34,[89][90][91] . The similar patterns of change detected following stress in HDAC4/5 may reflect that both of these can be regulated by the same microRNAs 92,93 . Although we show HDAC4/5 patterns to be similarly affected by stress and stress history, differential HDAC4/5 activity could arise through differential subcellular distribution, actions of histone acetyl transferases, or interaction with downstream targets 94,95 . Here, levels of an HDAC5 target, NPAS4, were not affected by stress manipulation, emphasizing the independent regulation of gene targets and suggesting that NPAS4 may not be a gene target consolidating effects of traumatic stress exposure.
A key extension of the current study would be the inclusion of female rodents. Epigenetic effects of developmental stress can be greater in females than in males 96,97 . Further, HDAC binding and expression in the brain is sex specific, and sex steroids modulate effects of stress on HDAC regulation 33,[98][99][100][101] . Additionally, humans and rodents show sex specificity in effects of traumatic stress and learned fear responses, including fear conditioning behavior [102][103][104] . A limitation of the current study is that effects of adolescent-stress are not isolated or directly compared to effects during adulthood or other, earlier developmental stages. Given the extensive maturational changes of systems investigated here, and unknown ontogenetic changes in HDAC4/5, effects of stress during other phases may differ in magnitude or direction. Finally, all neurochemical measures were obtained at a single time-point, such that plasticity is unclear. Time-course studies could determine whether adolescent-stress prevents change in HDAC4/5 following traumatic stress or reverses effects after a delay as well as define effects of stress timing and intensity.

Conclusion
Our results demonstrate a novel, unanticipated capacity for stress in adolescence to buffer effects of adult traumalike stress on a cognitive deficit characteristic of PTSD as well as HDAC4/5 and NE levels. Our findings also expand current models of developmental stress with respect to epigenetic regulation in adulthood, by demonstrating that HDAC4 and HDAC5 are dynamically regulated following stress exposure in a manner that reflects stress history. Given this result, and prior evidences that HDAC4/5 have key roles regulating effects of stress on behavior, complex cognition, glucocorticoids, and neural activity, evaluation of HDAC4/5 in the context of resilience and population subgrouping could be informative 29,35,36 . Additionally, our findings suggest that stress history can drive heterogeneity in responses to traumatic stress in adulthood through dynamic HDAC programming effects, which may manifest in clinical inconsistencies in the effects of childhood stress on vulnerability to adverse effects of adult trauma.