Early Life Stress Preceding Mild Pediatric Traumatic Brain Injury Increases Neuroinflammation but Does Not Exacerbate Impairment of Cognitive Flexibility during Adolescence

Abstract

Early life stress (ELS) followed by pediatric mild traumatic brain injury (mTBI) negatively impacts spatial learning and memory and increases microglial activation in adolescent rats, but whether the same paradigm negatively affects higher order executive function is not known. Hence, we utilized the attentional set-shifting test (AST) to evaluate executive function (cognitive flexibility) and to determine its relationship with neuroinflammation and hypothalamic–pituitary–adrenal (HPA) axis activity after pediatric mTBI in male rats. ELS was induced via maternal separation for 180 min per day (MS180) during the first 21 post-natal (P) days, while controls (CONT) were undisturbed. At P21, fully anesthetized rats received a mild controlled cortical impact (2.2 mm tissue deformation at 4 m/sec) or sham injury. AST was evaluated during adolescence on P35–P40 and cytokine expression and HPA activity were analyzed on P42. The data indicate that pediatric mTBI produced a significant reversal learning deficit on the AST versus sham (p < 0.05), but that the impairment was not exacerbated further by MS180. Additionally, ELS produced an overall elevation in set-loss errors on the AST, and increased hippocampal interleukin (IL)-1β expression after TBI. A significant correlation was observed in executive dysfunction and IL-1β expression in the ipsilateral pre-frontal cortex and hippocampus. Although the combination of ELS and pediatric mTBI did not worsen executive function beyond that of mTBI alone (p > 0.05), it did result in increased hippocampal neuroinflammation relative to mTBI (p < 0.05). These findings provide important insight into the susceptibility to incur alterations in cognitive and neuroimmune functioning after stress exposure and TBI during early life.

Introduction

Traumatic brain injury (TBI) in children represents a significant public health problem.¹ The burden is especially prominent in low- and middle-income countries where the risk factors for TBI are more prevalent.¹ The worldwide incidence of pediatric TBI ranges broadly and varies greatly by country, with most reporting a range between 47 and 280 per 100,000 children.² However, relative to the global rate, Latin American countries show higher TBI rates in all ages (150 per 100,000).¹ The increased incidence is noted particularly among children >3 years of age, where males show higher rates of TBI than females and mild TBI (mTBI) constitutes >80% of injuries.² Deficits in executive function and cognitive flexibility are among the most frequently reported cognitive impairments after pediatric TBI.^3–6 Executive function and cognitive flexibility are defined as the “individual's ability to navigate overlapping stimuli in the environment in order to unlearn a previously valid set of rules, filter unwanted distractions, and acquire a new rule by shifting attention from a salient stimulus dimension to a previously irrelevant one.”⁷ Typical development of these skills requires the recruitment of a widespread neural network of brain regions such as the pre-frontal cortex (PFC), parietal cortex, basal ganglia, and hippocampus.⁸

Survivors of non-accidental pediatric head trauma have significantly worse injuries and outcomes than those with accidental TBI.⁹ Although abusive head trauma (AHT) represents 38% of pediatric TBI, it is the cause of 71% of deaths and 90% of severe disability cases.¹⁰ According to the world health organization (WHO), child abuse and neglect are some of the “most intensive and frequently occurring sources of stress that children may suffer early in life.”¹¹ Moreover, early life stress (ELS) may increase the vulnerability to many neurological and behavioral disorders later in life.^12–14 Between 18% and 38% of pediatric TBI patients exhibit executive dysfunction during the 1st year after injury, which strongly suggests that factors commonly associated with ELS, such as learning and behavior problems, limited family resources, and poor family functioning adversely affect executive function after TBI.³ Moreover, ELS can lead to structural and functional impairments in the PFC,^15,16 and such changes in neural circuits supporting executive impairments caused by ELS could exacerbate the sequelae of pediatric TBI. However, despite the plethora of information indicating that AHT differs substantially from accidental brain injuries, most pre-clinical pediatric TBI studies do not consider the possible interaction between ELS and TBI. Hence, translation of novel interventions from bench to bedside could be hampered by the lack of studies that justly represent the complexity of abuse, stress, and head trauma relationships observed in victims of AHT.

It has been shown that ELS activates the hypothalamus–pituitary–adrenal (HPA) axis and increases stress reactivity in adulthood in human¹⁷ and animal studies.^18–22 One potential consequence of increased HPA axis activity after ELS is hypersensitivity to secondary challenges (double hit), including subsequent injuries, stressors, and infections.^19,23–25 In pre-clinical models of ELS, activation of the neonate HPA axis response corresponds with the development of a primed and more activated microglia phenotype and increased hippocampal levels of proinflammatory cytokines, which are also increasingly touted as biomarkers for poor prognosis after TBI,²⁶ such as interleukin 1 beta (IL-1β), interleukin 6 (IL-6), and tumor necrosis factor alpha (TNF-α).^25,27,28 Increased HPA axis activity and inflammatory cytokine production from primed/activated microglia could impair normal neurological function in the developing brain.^18,29–34 Moreover, during the early post-injury period, TBI activates the HPA axis response and causes an exacerbated inflammatory response that is mediated by these same cytokines.^35,36 Acute inflammation is a homeostatic mechanism that facilitates the removal of cellular debris and tissue repair processes. However, inadequate resolution of acute inflammation amplifies the injury.³⁷ Despite increasing evidence on the effects of TBI on neuroendocrine-immune function during the early phase, the responses of this system at later time points are poorly understood.

Maternal separation (MS180) is a widely used rodent model of ELS that mimics many of the long-term consequences observed in victims of abuse.^18-20,38-40 MS180 activates the HPA axis response after each separation episode even during the stress hypo-responsive period, and affects brain development.^{18,27,38,40,41} Previous work from our group demonstrated that MS180 in combination with a pediatric mTBI causes a deficiency in cognitive performance in a spatial learning task that is accompanied by an increase in microglial activation in the area around the injury and hippocampus in the chronic phase of TBI.³⁷ Thus, an increased vulnerability to mTBI in stressed animals could be possibly linked to a failure in the resolution of inflammatory processes. However, the effect of pediatric mTBI, alone or in combination with a previous exposure to ELS, on cognitive flexibility and proinflammatory mediators such as cytokines, remains unknown. The aim of the current study is to evaluate the effects of pediatric mTBI on cognitive flexibility. Additionally, we seek to investigate how ELS-induced HPA axis programming and neuroinflammation might contribute to influencing further the effects of pediatric mTBI.

Methods

Rats

Ten time-pregnant female Sprague–Dawley rats (Envigo RMS, Inc., Indianapolis, IN) were obtained on embryonic day 14. The rats were maintained in specific pathogen-free temperature-controlled rooms with a 12-h light/dark cycle and ad libitum food and water. Day of birth was considered as postnatal day zero (P0). On P1, litters were cross-fostered, culled to eight pups (four to six males), and randomly assigned to the MS180 or control (CONT) groups. Pups were weaned at P21 and received a mild controlled cortical impact or sham injury. To avoid sex and litter-dependent effects, only males were evaluated, and each experimental group consisted of pups from at least five different dams (n = 8 per group). All experiments and manipulations, including euthanasia, were performed between 9:00 a.m. and 12:30 p.m. All procedures were approved by the University of Pittsburgh's Institutional Animal Care and Use Committee and the National Research Committee of the Mexican Social Security System (R-2016-785-054). All experiments were conducted in accordance with the Institute of Laboratory Animal Resources Guide for the Care and Use of Laboratory Animals (National Research Council, revised 1996) and the official regulations for use and care of laboratory animals of Mexico (NOM-062-ZOO-1999). Every effort was made to reduce the number of rats used, and to minimize suffering.

Maternal separation

MS180 was performed as previously described.^18-20,42 Briefly, on P1, MS180 pups were separated from their dam and placed in a box filled with clean sawdust and relocated in a room separate from the colony, with controlled temperature (30–32°C) for 3 h and then returned to the dam. This procedure was repeated every day at the same time (9:00 a.m. to 12:30 p.m.) until weaning at P21. The CONT group remained undisturbed except for routine cage cleaning, which was very brief. To further avoid potential confounding effects of handling and manipulation, body weight was not evaluated during the pre-weaning period.

Controlled cortical impact (CCI) injury

Rats weighing 47–65 g were subjected to a CCI or sham injury on P21 as previously described.⁴² Briefly, anesthesia was induced and maintained with inspired concentrations of 4% and 2% isoflurane, respectively, in 2:1 N₂O:O₂. The rats were secured in a stereotaxic frame, and a midline scalp incision was made under aseptic conditions. The skin and fascia were reflected to expose the skull, and a craniectomy, encompassing bregma and lambda and the sagittal and coronal sutures, was made in the right hemisphere with a dental drill. The bone flap was removed, the impacting rod was extended, and the impact tip (6 mm, flat) was centered and lowered through the craniectomy until it touched the dura mater. Then the rod was retracted, and the impact tip was advanced 2.2 mm farther to produce a mild TBI (2.2 mm tissue deformation at 4 m/sec). Body temperature was monitored with a rectal probe and maintained at 37 ± 0.5°C with a heating blanket. Immediately after the CCI, anesthesia was discontinued, and the incision was promptly sutured. Sham-operated rats underwent similar surgical procedures but were not subjected to the impact. Post-injury weight gain was monitored daily from P21 until euthanasia at P42. Every consideration was taken to minimize stress during the weighing procedure.

Acute neurological evaluation

Hindlimb reflexive ability was assessed immediately following the cessation of anesthesia and removal from the stereotaxic apparatus by gently squeezing the rats' paw every 5 sec and recording the time to elicit a withdrawal response. Return of the righting reflex was determined by the time required to turn from the supine to the prone position. These sensitive neurological indices are used to determine the level of injury severity.^7,42

Attentional set-shifting test (AST)

The AST task^43,44 was adapted from Birrell and Brown⁴³ and implemented after TBI as described previously.^45–47 Briefly, 1 week prior to behavioral testing (i.e., day 7 post-surgery), rat food intake was mildly restricted to 12 g per day, consisting of 80% of the rat's normal diet, with water freely available. Rats were first trained to dig reliably in small terracotta pots (internal rim diameter, 7 cm; depth, 6 cm) to obtain the food reward in clean home cages and then further trained and tested on the subsequent 2 days in the testing arena, which were custom-built rectangular Plexiglas boxes (30 × 51 × 25 cm) painted gray on the outer side of all surfaces. Each pot was scented with a single odor by adding two drops (10 μL/drop) of aromatic oil (NOW Foods, Bloomingdale, IL) to the inner rim 5 days prior to use. Odors were reapplied daily (1–2 μL) to maintain consistent intensity. A different pot was used for each combination of digging medium and odor, and only one odor was ever applied to a given pot. The “reward” was one quarter of a Honey Nut Cheerio (General Mills Cereals, Minneapolis, MN) buried 2 cm below the surface of the digging medium in the “positive” pot. To avoid location by smell rather than learning and discrimination, pots were sprinkled with Cheerio dust prior to each stage.

For the training phase, rats were trained on a series of simple discriminations (SD) to reach a criterion of six consecutive correct trials. First, they had to learn to discriminate between aromatic odors in sawdust-filled pots (lemon vs. eucalyptus). After reaching criterion for the odor discrimination, rats learned to associate the food reward with a medium cue using new unscented pots (felt vs. paper strips). All rats were trained using the same pairs of stimuli and in the same order. On the following day, rats were then tested on a series of increasingly difficult discrimination stages, each also requiring a criterion of six consecutive correct trials, based on the cue progression previously established.^45,46

The first stage was an SD involving only one stimulus dimension, with half of the rats starting with odor and half starting with medium (for clarity, subsequent description will consider only the example beginning with an odor discrimination; the stimulus shift performance is also analyzed to ensure there is no perceptual dimension bias affecting AST results on test day). The second stage was a compound discrimination (CD), in which the same contingency rule was required (e.g., odor), and the second, irrelevant stimulus dimension (e.g., medium) was introduced. As in the SD task, only one odor was associated with reward, but two different digging media were paired randomly with the odors. The third stage was a reversal (R1) of the previous discrimination, in which the same odors and media were used. Odor was maintained as the relevant dimension; however, the negative odor from the previous stage became positive and the positive odor from the previous stage became negative (no reward). The fourth stage was an intradimensional (ID) shift in which all new stimuli (odors and media) were introduced. Again, odor remained the relevant dimension and digging medium was still irrelevant. The fifth stage was a reversal of this discrimination (R2), in which the previously negative odor became positive, similar to R1. The sixth task required an extradimensional (ED) cognitive set-shift, in which all new stimuli were again introduced, but the dimension that had been repeatedly reinforced as the informative, relevant dimension (thus forming a “cognitive set”) was now irrelevant, and the previously distracting dimension (i.e., the digging medium) became the relevant dimension. Finally, the seventh stage was another reversal (R3), where the previously negative stimulus became positive, as in the previous reversals. The rats were allowed 10 min to make a choice on each trial. If a choice was not made within this interval, the trial was scored as an error and the rat was returned to the start box. Rats failing to make a choice on 6 consecutive trials, or failing to complete a stage within 50 trials, were eliminated from further testing. The dependent measures were the number of trials required to reach criterion for each stage of the test, the total number of errors per stage, as well as set-loss errors, which occurred when 50% of the rule contingency had been achieved (i.e., three, four, or five correct responses followed by an incorrect choice).⁴²

Tissue preparation, corticosterone assay, and quantitative polymerase chain reaction (qPCR)

On P42, the rats were transported from the main housing room to the local vivarium at 9:00 a.m. and were left undisturbed for 90 min to avoid the confounding effects of transport stress. After the 90 min, the rats were transported to the necropsy room one at a time, given an overdose of sodium pentobarbital (Nembutal, 80–100 mg/kg; Abbott Laboratories, North Chicago, IL), and then were decapitated and trunk blood was collected in tubes containing aprotinin and EDTA (pH 8, Sigma-Aldrich, St Louis, MO, USA) to analyze plasma corticosterone (CORT) levels. The PFC, hippocampus (ipsilateral and contralateral), and hypothalamus were dissected on a chilled ice plate to analyze the expression of cytokines and HPA axis related neuropeptides as previously described.⁴³ Brain tissues were placed in Trizol Reagent (Invitrogen, Camarillo, CA), and stored at -70°C until RNA extraction was performed following the manufacturer's instructions. Plasma was stored at -70°C until processing for CORT analysis by ELISA (ENZO life sciences, Cat. No: ADI-900-097) and processed according to the manufacturers protocol for small volumes of plasma samples to assure that all samples fell within the essay detection range (0.032–20 ng/mL). Briefly, 10 μL of each sample were pre-treated with a 10 μL of 1:100 steroid displacement reagent (SDR) solution (immediately before use) for 5 min and further diluted in 380 μL of assay buffer to a final dilution of 1:40.

Real time (RT)-qPCR

We used RT-qPCR to determine cytokine expression (IL-1β, IL-6, and TNF-α) in the hippocampus and the PFC, as well as the expression of HPA axis activity-related genes corticotropin-releasing hormone (CRH) and arginine vasopressin (AVP) in hypothalamic tissue as previously described^27,48 (see Table S1 for primer sequence and insertion number). Hypoxanthine-guanine phosphoribosyltransferase (Hprt) was used as the reference housekeeping gene. Total RNA from hippocampal and hypothalamic tissues was extracted using Trizol reagent (Invitrogen, Camarillo, CA) following manufacturer instructions. A total of 1 μg of RNA per sample was subjected to reverse transcription in a 20-μL reaction solution containing 1 μL of Oligo dT (Promega, 500 μg/mL), dNTP mix (dATP, dCTP, dGTP, and dTTP, 10mM each; Invitrogen, Camarillo, CA), first strand buffer, and 200 U of M-MLV reverse transcriptase (Promega, 200U/1μL), at 37°C for 50 min. We used 250 ng of cDNA for RT-qPCR analyses with each reaction containing Master Mix 2X (iTaq Universal SYBR Green Supermix, BioRad) and forward and reverse primers (10 μM/μL each, Invitrogen, Camarillo, CA). RT-qPCR was performed for 40 cycles using the CFX 96 Touch Real-Time PCR Detection System (Bio-Rad). Each sample was run in duplicate. Data were normalized and analyzed using the ΔΔC_T method adapted from Bustin.⁴⁹ All experiments were performed in accordance with the MIQE Guidelines: Minimum Information for Publication of Quantitative Real-Time PCR Experiments.⁵⁰ Intra- and inter-assay coefficients of variation were <10% for all comparisons.

Statistical analysis

Statistical analyses were performed on data generated by experimenters blinded to conditions using GraphPad Prism 7.0 (GraphPad software, US). Post mTBI body weight gain and cognitive test day outcomes were analyzed by three-way repeated-measures analysis of variance (ANOVA) with ELS, injury, and stage as factors, of which stage was the repeated measure. AST training and ED shift type performance were analyzed by two-way repeated-measures ANOVA with stage as the repeated measure. The acute neurological assessment, CRH, AVP, and CORT levels were analyzed by two-way ANOVAs with ELS and injury as factors. When the ANOVA revealed a significant effect, Tukey's post-hoc tests were utilized to determine specific group differences. Cytokine expression was analyzed with a three-way ANOVA considering ELS, TBI, and injury side as factors. No differences were observed between ipsilateral and contralateral cytokine expression in sham rats, and therefore, the data were pooled and further analyzed with a one-way ANOVA. The results are expressed as the mean ± standard error of the mean (SEM), and were considered significant when p ≤ 0.05.

For correlation analysis, entire data sets for each variable (i.e., all rats) were tested with the Shapiro–Wilk test for normal distribution and a Pearson's correlation analysis was estimated using GraphPad Prism 7.0 (GraphPad software, US). A heat map was obtained for Pearson's correlation coefficients and p values. Stepwise multivariate linear regression models were performed to confirm correlations and predict performance in the AST based on individual biomarkers (ipsilateral and contralateral cytokine expression and HPA axis activity) and injury severity with the Statistical Package for the Social Sciences (SPSS) software.

Results

Body weight gain and acute neurological evaluation

A Student's t test for pre-surgery weights indicated that there were no differences between CONT and MS180 rat pups at P21 (p > 0.05). Three-way repeated measures ANOVA for post-mTBI body weight gain indicated a significant effect of age (F_20,678 = 145.5, p < 0.00001), but no effect of mTBI or MS180 (p > 0.05). Two-way ANOVAs showed significant effects of mTBI for return of hindlimb reflex ability after a brief paw pinch (right, F_1,28 = 317.8, p < 0.00001 and left, F_1,28 = 322.3, p < 0.00001) and righting reflex latency (F_1,28 = 113.0, p < 0.00001) following the cessation of anesthesia (Table 1). Specifically, both mTBI groups took longer to gain reflexive and righting ability relatively to sham controls (p < 0.05) but did not differ from one another (p > 0.05).

Table 1.

Mild Pediatric TBI Affects Acute Neurological Evaluation

	CONT + sham	MS180 + sham	CONT + TBI	MS180 + TBI
Right reflex (s)	8.6 ± 2.8	7.3 ± 3.7	129 ± 36.7***	136.8 ± 14.2***
Left reflex (s)	12.5 ± 3.1	11.1 ± 3.7	132.7 ± 36.2***	140.2 ± 14.4***
Righting reflex (s)	84.6 ± 20	72.8 ± 10.1	256.2 ± 86.4***	251.9 ± 26.9***

Control (CONT) and maternally separated (MS180) rats were subjected to either a mild traumatic brain injury (mTBI) or a sham injury on P21. Hindlimb reflexive ability and return of the righting reflex was determined following the cessation of anesthesia. Values are mean ± standard error of the mean (SEM). Analysis of variance (ANOVA) ^***p < 0.0001 versus CONT + sham and MS180 + sham, which did not differ from one another. CONT + Sham (n = 8), MS180 + sham (n = 8), CONT + TBI (n = 8), and MS180 + TBI (n = 8).

Mild pediatric TBI impairs performance on the AST

There was no significant difference among groups for the number of trials required to reach criterion for the SD tasks on the training day (F_1,28 = 2.419; p = 0.13). There was also no effect of mTBI (F_1,28 = 2.783, p = 0.106) or MS180 (F_1,28 = 0.003, p = 0.957) on the number of trials required to reach criterion for the SD tasks on the training day, nor was there a significant interaction. A two-way repeated measures ANOVA was performed for the comparison between rats subjected to an odor-to-medium versus those subjected to a medium-to-odor ED set-shift on the trials to reach the criterion for each task of the AST regardless of injury or treatment group. There was a significant main effect of ED shift type (F_1,30 = 6.328, p = 0.017) on AST performance across stages, as well as a significant ED shift type x stage interaction (F_6,180 = 3.01, p = 0.008). Tukey post-hoc analyses revealed a significant difference on CD (p < 0.05), suggesting that rats subjected to an odor-to-medium relevant dimension switch required more trials to criterion than rats subjected to a medium-to-odor switch. No other stage comparisons were significantly different based of the dimension to which the contingency belonged, such as odor or medium.

Three-way repeated measures ANOVA for trials to criterion on AST revealed a main effect of stage (F_6,168 = 5.951, p = 0.00001) and injury x stage interaction (F_6,168 = 2.274, p = 0.039), but not of injury or maternal separation alone (p > 0.05). Post-hoc analyses for the main effect of stage showed that when collapsing all groups together, rats performed more trials on R1 regardless of group condition compared with SD, CD, ID, and R2 (p < 0.05) and also produced more errors in R2 relative to SD, therefore suggesting that reversal learning was generally more challenging and validating the inherent difficulty built within the task sequence.⁴⁴ Subsequent post-hoc two-way ANOVAs (injury x MS) for individual AST stages revealed a significant injury effect (F_1,28 = 16.72, p = 0.0003) on the first reversal stage (R1), but no effect of separation or interaction (p > 0.05). Post-hoc comparisons indicated that both CONT + TBI and MS180 + TBI rats showed an increase in the number of trials to reach criterion during this stage when compared with CONT + sham group (p < 0.05, Fig. 1). No differences were observed on any of the other stages of testing. Given the TBI-related impairment on R1, we also performed a two-way ANOVA on CD (injury x ED shift type) to ensure that the rats' propensity to require more trials to criterion in this stage did not influence performance on the subsequent stage. There was not a significant effect of either injury (p > 0.05), or injury x ED shift interaction (p > 0.05), demonstrating that both TBI and sham rats had an equally harder time on CD when they commenced AST with odor as the relevant dimension, which was independent of the TBI-induced deficit in reversal learning.

FIG. 1.

Mild pediatric TBI caused significant impairments in the first reversal stage of the attentional set-shifting test (AST). (A) Control (CONT) and maternally separated (MS180) rats were subjected to either a mild traumatic brain injury (mTBI) or a sham injury on P21. Rats recovered for 5 days and then were further food restricted from P27 through P34. Executive function in the AST was evaluated from P35 through P40. (B) Number of trials required to reach criterion of six consecutive correct trials in the different stages of the AST. Data are expressed as mean trials to criterion + standard error of the mean (SEM). TBI, traumatic brain injury; SD, simple discrimination; CD, compound discrimination; ID, intradimensional ED, extradimensional; R1, first reversal; R2, second reversal; R3, third reversal; AST, attentional set-shifting test. (Analysis of variance [ANOVA] **p < 0.01 vs. CONT + sham). CONT + sham (n = 8), MS180 + sham (n = 8), CONT + TBI (n = 8), and MS180 + TBI (n = 8).

A three-way repeated measures ANOVA for the total number of errors per stage showed a main effect of stage (F_6,168 = 8.229, p < 0.00001), but not of injury or maternal separation alone, as well as no interactions (p > 0.05); therefore, no group-related inferences can be made on this measure. Further, a three-way repeated measures ANOVA for the total number of set-loss errors on each stage revealed a significant effect of maternal separation (F_1,28 = 4.755, p = 0.038), but not of injury or stage alone, as well as no interactions (p > 0.05). Post-hoc analyses for the separation factor did not render effects on individual test stages, suggesting that MS180 induced an overall elevation in set-loss errors across the AST, seen as an increased propensity to make mistakes after ≥50% of the rule contingency has been achieved; that is, reduced ability to maintain acquisition of correct stimulus contingency.

MS180 increased corticosterone levels

Two-way ANOVA showed there was no effect of mTBI, MS, or their interaction on hypothalamic CRH expression (p > 0.05, Fig. 2A). There was a significant main effect of mTBI (F_1,28 = 5.5, p = 0.026), but not of MS or interaction (p > 0.05, Fig. 2B) on hypothalamic AVP expression; however, multiple comparisons failed to show further differences between the groups. Statistical analysis for trunk blood corticosterone levels showed a significant effect of MS (F_1,28 = 10.9, p = 0.0026), but not of mTBI or interaction (p > 0.05, Fig. 2C). Multiple comparisons indicated that th e MS180 + sham and MS180 + mTBI groups had a significant increase in CORT plasma levels when compared with CONT + sham group (p < 0.05).

FIG. 2.

Early life stress increases glucocorticoid levels in adolescent rats. Control (CONT) and maternally separated (MS180) rats were subjected to either a mild traumatic brain injury (mTBI) or a sham injury on P21. Rats were euthanized on P42. Relative expression (fold change) of hypothalamic (A) corticotropin-releasing hormone (CRH) and (B) arginine vasopressin (AVP) was quantified by quantitative polymerase chain reaction (qPCR). (C) Corticosterone (CORT) concentration in trunk blood was quantified by ELISA. Dotted lines depict biologically significant fold change increase in the qPCR. (Analysis of variance [ANOVA] *p < 0.05 vs. CONT + sham). CONT + sham (n = 8), MS180 + sham (n = 8), CONT + TBI (n = 8), and MS180 + TBI (n = 8).

ELS increases hippocampal IL-1β after mTBI

A three-way ANOVA was performed to analyze the effect of treatment and injury side on IL-1β, IL-6, and TNF-α expression in the hippocampus. We observed a significant effect of injury side, mTBI, and injury side x mTBI x MS interaction (Table S2) on hippocampus IL-1β expression. No differences were observed between ipsilateral and contralateral cytokine expression in sham rats, and therefore, the data were pooled. Statistical analysis (one-way ANOVA) indicated a significant group effect on hippocampal IL-Iβ expression (F_5,40 = 7.83, p = 0.00003) and multiple post-hoc comparisons showed that MS180 + mTBI rats exhibited a significant increase in IL-1β expression in the ipsilateral hemisphere than the sham (p < 0.001) and CONT + mTBI (p < 0.05) groups (Fig. 3A). A significant difference between CONT + sham and CONT + mTBI on IL-1β expression was observed only when performing a single Student t test (p = 0.02). There was no effect of treatments (p > 0.05) in hippocampal IL-6 and TNF-α (Fig. 3B and C).

FIG. 3.

Early life stress increases hippocampal interleukin (IL)-1β expression after traumatic brain injury (TBI). Control (CONT) and maternally separated (MS180) rats were subjected to either a mild TBI (mTBI) or sham injury on P21. Rats were euthanized on P42. Relative expression (fold change) of (A) IL-1β, (B) IL-6 and (C) tumor necrosis factor alpha (TNF-α) was quantified by quantitative polymerase chain reaction (qPCR). Dotted lines depict biologically significant fold change increase. (Analysis of variance [ANOVA] ***p < 0.001 vs. CONT + sham; ^###p < 0.001 vs. CONT + TBI; ^$p < 0.05 vs. CONT + TBI ipsilateral). CONT + sham (n = 8), MS180 + sham (n = 8), CONT + TBI (n = 7), and MS180 + TBI (n = 8).

Three-way ANOVAS (Table S2) for PFC IL-1β and TNF-α expression indicated that there was a significant effect of injury side and injury side x mTBI interaction on IL-1β expression and of injury side and injury side x mTBI x MS interaction on TNF-α expression. Multiple comparisons indicated that MS180 + TBI significantly increased IL-1β PFC expression when compared with the contralateral side of the CONT + mTBI (p < 0.05) and their own contralateral side (p < 0.05); however, no differences were observed when compared with sham groups. No differences where observed between ipsilateral and contralateral cytokine expression in sham rats and, therefore, the data were pooled. One-way ANOVA for pooled data indicated there was no further effect of treatments in PFC cytokine expression (p > 0.05, Table 2).

Table 2.

Early Life Stress and mTBI Do Not Affect PFC Cytokine Expression

	CONT + sham	MS180 + sham	CONT + mTBI contralateral	MS180 + mTBI contralateral	CONT + mTBI ipsilateral	MS180 + mTBI ipsilateral
IL-1β	1 ± 0.25	0.82 ± 0.19	0.60 ± 0.26	1.03 ± 0.51	0.86 ± 0.29	1.80 ± 0.43
IL-6	1 ± 0.37	1.32 ± 0.31	5.03 ± 3.23	1.15 ± 0.32	1.76 ± 0.41	2.07 ± 0.36
TNF-α	1 ± 0.53	1.62 ± 0.74	2.5 ± 1.60	1.51 ± 0.67	0.76 ± 0.32	2.75 ± 1.37

Control (CONT) and maternally separated (MS180) rats were subjected to either a mild traumatic brain injury (mTBI) or sham injury on P 21. Rats were euthanized on P42, the pre-frontal cortex (PFC) were dissected and interleukin- 1beta (IL1β), interleukin- 6 (IL-6) and tumor necrosis factor alpha (TNF-α) relative expression (fold change) was quantified by quantitative polymerase chain reaction (qPCR). Values are mean ± standard error of the mean (SEM). CONT + sham (n = 8), MS180 + sham (n = 8), CONT + TBI (n = 7), and MS180 + TBI (n = 8).

Hippocampus IL-1β expression predicts executive function

Entire data sets were analyzed with a Pearson's correlation, and a stepwise multivariate linear regression model was applied. No correlations were observed for the stages of the AST (Fig. 4A cluster I). As expected, significant correlations were observed for intra-structure and intra-hemispheric inflammatory markers in the contralateral hemisphere (Fig. 4A cluster II). This intra-structure and intra-hemispheric correlation pattern was disturbed in the ipsilateral side (Fig. 4A cluster III). No correlation was observed between HPA axis activity markers (Fig. 4A cluster IV). Also, as we anticipated, a strong correlation was observed between all the acute neurological assessment variables (Fig. 4A cluster V).

FIG. 4.

Executive function impairments correlate with interleukin (IL)-1β expression. (A) Color heat map of the Pearson's correlation coefficients computed for the number of trials to reach criterion in each of the phases of the attentional set-shifting test (AST) (cluster I), individual inflammation biomarkers (cluster II: contralateral and cluster III: ipsilateral cytokine expression), hypothalamic–pituitary–adrenal (HPA) axis activity (cluster IV), and injury severity (neurological assessment scores, cluster V). Correlation coefficients are shown with continuous gradient colors, where significant positive correlation is marked in red (r > 0.3, *p < 0.05, **p < 0.01), negative is marked in blue (r < −0.3, *p < 0.05, **p < 0.01), and white represents no significant correlation (p > 0.05). (B) Scatterplot graph for stepwise multivariate linear regression model indicating that IL-1β expression in the ipsilateral hippocampus predicts for the number of trials to reach criterion on the R1 of the AST. SD, simple discrimination; CD, compound discrimination; ID, intradimensional ED, extradimensional; R1, first reversal; R2, second reversal; R3, third reversal; TNF, tumor necrosis factor; PFC, pre-frontal cortex; CRH, corticotropin-releasing hormone; AVP, arginine vasopressin; CORT, corticosterone.

A significant correlation was observed for cognitive deficits (number of trials to reach criterion in the R1 phase) and IL-1β expression in both the ipsilateral PFC (r₂₉ = 0.491, p = 0.005) and the hippocampus (r₂₉ = 0.425, p = 0.036), as well as TNF-α expression in the ipsilateral PFC (r₂₉ = 0.425, p = 0.017). Performance on the CD phase of the AST showed significant correlations with cytokine expression in the contralateral hippocampus (IL-1β r₂₉ = 0.450, p = 0.011; IL-6 r₂₉ = 0.443, p < 0.012; and TNF-α r₂₉ = 0.482, p = 0.006). The number of trials to reach criterion on the CD showed an additional correlation with TNF-α expression in the ipsilateral PFC (r₂₉ = 0.412, p = 0.021). The number of trials to reach criterion on the ID of the AST positively correlated with IL-1β expression in the contralateral (r₂₉ = 0.407, p = 0.023) and ipsilateral PFC (r₂₉ = 0.367, p = 0.042). CRH expression showed a significant positive correlation with hypothalamic IL-1β expression in the ipsilateral PFC (r₂₉ = 0.663, p = 0.0001) and a negative correlation with hippocampal IL-6 expression in the ipsilateral side (r₂₉ = -0.394, p = 0.035). IL-1β expression in the ipsilateral hippocampus showed additional correlations with AVP (r₂₉ = 0.456, p = 0.019) and injury severity (right hindlimb reflex r₂₉ = 0.492, p = 0.005, left hindlimb reflex r₂₉ = 0.491, p = 0.005, righting reflex r₂₉ = 0.458, p = 0.010). CORT plasma levels showed a positive correlation with injury severity (righting reflex r₃₀ = 0.391, p = 0.027).

A stepwise multivariate linear regression model was calculated to predict performance in the first reversal trial of the AST based on individual biomarkers (ipsilateral and contralateral cytokine expression, HPA axis activity) and injury severity. Stepwise multivariate linear regression model confirmed that only IL-1β expression in the ipsilateral hippocampus significantly predicted for cognitive performance in the AST (F_1,16 = 12.07, p = 0.003) with an R of 0.66 (Fig. 4B). The predicted number of trials to reach criterion is equal to 0.04 (IL-1β) + 0.78, where IL-1β is measured as fold change with respect to sham + CONT expression in the contralateral hemisphere.

Discussion

In the present study we evaluated for the first time the effect of pediatric mTBI, alone or preceded by ELS on executive function, pro- inflammatory cytokines, and HPA axis activity during adolescence. The data showed that pediatric mTBI caused a significant impairment in the first reversal stage of the AST, indicating cognitive inflexibility, which correlates with IL1-β expression in the hippocampus–PFC circuit. Multivariate stepwise linear regression analysis further confirmed that IL-1β expression in the ipsilateral hippocampus significantly predicts for cognitive performance in the R1 of the AST. The data further showed that although ELS increases IL-1β expression after mTBI, it does not further exacerbate mTBI-induced impairment in the AST. Moreover, ELS, but not TBI, affected HPA axis activity.

AST is a behavioral test that has been validated as a sensitive tool to assess cognitive flexibility deficits induced by TBI in adult rats.^45,46 It has been shown that moderate-severe pediatric TBI induced at P12 impaired reversal learning in a visual discrimination task on a touch-screen operant chamber platform.⁵¹ Evaluating the effect of pediatric mTBI on executive function using the AST is novel and highly warranted in order to advance knowledge about the cognitive domains affected in this TBI model. In adults, moderate and severe TBI (2.8 and 3.0 mm cortical tissue deformation, respectively) caused significant cognitive impairments in the first reversal of the AST, the ED stage, as well as on the third reversal stage compared with sham controls; however, mild TBI (2.6 mm tissue deformation) was not severe enough to induce cognitive alterations in this test.⁴⁵ Conversely, we observed that a milder injury (2.2 mm tissue deformation) is robust enough to affect cognitive flexibility when applied in pediatrics, indicating that the developing brain is more sensitive to TBI-induced impairments than the adult. It has also been shown that MS causes cognitive deficits in the CD, R1, and ED stages of the AST in adult rats;⁵² however, this effect was not present in adolescents on individual AST stages, as MS180 + sham rats do not show a significant difference in trials to reach criterion compared with CONT + sham. Nevertheless, MS induced an overall increase in total set-loss errors on the AST, suggesting reduced ability to maintain the acquisition of correct stimulus contingency in ELS rats. Although a singular ED shift type effect on CD was unveiled, further post-hoc analyses determined that it was not biased to TBI or MS rats and it did not affect performance on subsequent, more complex stages. Rodents have stereo odor localization (i.e., they are macrosmatic). Ablations of telencephalon regions (olfactory bulb, anterior olfactory nucleus, primary olfactory cortex) or the olfactory-hippocampal pathway do not seem to interfere with scent discrimination capabilities in rats. Moreover, we apply aromatic oil scents in small amounts on stimulus pots, but above the human olfactory threshold; therefore, any post-TBI olfactory dysfunction in rats may not be significant enough to dramatically interfere with AST performance.⁴⁵

Studies of HPA axis activity after TBI are conflicting. Some groups have reported that TBI can cause an endocrine dysfunction during the chronic phase of TBI that decreases corticosterone, CRH, and AVP,^53–55 whereas others indicate that TBI increases HPA axis activity.^51,56,57 In the present study, we did not observe an effect of mTBI on HPA axis activity markers at euthanasia. Because the mTBI model utilized here was selected so that it would not cause a ceiling effect that hampered our ability to detect the contribution of previous exposure to ELS, it is plausible that the intensity of the injury was not severe enough to cause diffuse damage to the hypothalamic regions; however, this is unlikely, as we have previously observed that moderate-severe TBI in adults has no effect on corticosterone levels during the chronic phase of TBI.⁵⁸ It is important to note that CORT levels at euthanasia cannot be considered to be true baseline conditions, as they could be biased by the effect of anesthesia during euthanasia (stressor)⁵⁹ and hence, further experiments should be done utilizing non-stressing sampling techniques, like jugular vein catheterization, to determine real stress-free baseline CORT concentration in MS180 and TBI rats. As expected, in the present work we observed that ELS caused an elevation of CORT levels at euthanasia. The long-term effect of MS180 on HPA axis activity has been widely studied;^{18–20,48,60–62} nevertheless, reports on the effect of MS180 on adolescent HPA axis activity are scarce. Most of the studies on maternally separated adolescent animals have been performed in Wistar rats, where separation for 6 h a day (MS360) from P1 to P21 has been shown to increase CORT levels when compared with a handled-only group (MS15);⁶³ however, shorter separation periods were not able to program HPA axis activity in this strain.^64–66 It has been proposed that gene–environment interactions for each strain as well as individual resilience in outbred and inbred strains could account for many of the discrepancies observed in HPA axis programming effects of MS.^67,68

An unexpected result in our study was the lack of correlation among CRH, AVP, and CORT. This lack of direct correlation between the expression of hypothalamic neuropeptides and plasma CORT has been previously described in adult MS180 rats.^48,69,70 One of the limitations of our work is that we did not include adrenocorticotropic hormone (ACTH) quantification as one of the variables, and therefore, it is unknown if the absence of correlation between AVP and CRH with CORT levels is caused by a downregulation of CRH and AVP receptors at the pituitary level, or simply by differences in the methodology used to measure the variables (e.g., gene expression vs. plasma protein quantification). The endocrine response of the HPA axis is faster than the induction of gene expression. Therefore, we could hypothesize that the lack of correlation among CRH, AVP, and CORT levels could be because, as we mentioned before, CORT levels at euthanasia are elevated by the effect of anesthesia,⁵⁹ whereas the expression of CRH and AVP mRNA would require a longer period to be affected.

IL-1β, IL-6, and TNF-α are some of the most widely studied cytokines in the brain.⁷¹ IL-1β, IL-6, and TNF-α are strongly expressed in the hippocampus and PFC⁷² and have been proposed to modulate cognitive processes after TBI.^71,73–75 Here, we observed that IL-6 and TNF-α expression levels were not affected 21 days post-injury. These findings are similar to previous results indicating that cytokine levels normalize 48 h after a pediatric mTBI.⁷⁶ A significant finding was that MS180 + TBI rats showed a significant increase in hippocampus IL-1β expression, whereas a significant difference between CONT + sham and CONT + TBI on IL-1β expression was observed only when performing a single Student's t test. IL-1β can be produced by glial cells and it has been suggested that it plays a physiological role in modulating hippocampus synaptic plasticity.⁷⁷ In support, we have previously reported that MS180 + TBI increases hippocampal microglial activation after mTBI,⁴² suggesting that these activated cells could be responsible for the increase in IL-1β expression observed here. Moreover, IL-1β can function both as a pro-inflammatory and as an anti-inflammatory cytokine,⁷⁵ in such a way that TBI and ELS could modify the balance between these opposite qualities of IL-1β and cause a switch from its anti-inflammatory (M2) profile toward a pro-inflammatory one (M1).

In the present work, we performed a correlation analysis that allowed us to establish if the evaluated parameters could predict the cognitive deficits observed in the AST. As expected, Pearson's analysis showed that there is a correlation between the expression of hippocampal cytokines with those of PFC, which is altered on the side ipsilateral to the injury. Other important correlations were observed for performance in CD and several of the cytokines, mainly on the contralateral hippocampus, which suggests that these biomarkers could be involved in modulating compound discrimination in the healthy brain, thus fostering the ability to focus on overlapping stimuli and filter out distractors. The most significant correlation was that IL-1β expression in the hippocampus and PFC positively correlate with performance in the R1 of the AST. Additionally, IL-1β also showed a positive correlation with the severity of the lesion, suggesting that cognitive deficits caused by TBI could be mediated by the increase in the expression of this cytokine. To test this hypothesis, we performed a multivariate linear regression analysis, which confirmed that only IL-1β expression in the ipsilateral hippocampus predicts performance on R1 of AST. Further, we observed that, although stress exposure and TBI did not cause an increase in the hypothalamic expression of CRH, there was a positive correlation between the expression of IL-1β in the ipsilateral PFC and the expression of this neuropeptide. There is evidence that suggests that CRH could be an important mediator of the effects of ELS and TBI.^77–80 Hence, it is possible that both TBI and ELS could modify local CRH activity and modulate inflammatory tone.

We have previously reported that MS + TBI, but not pediatric TBI or MS180 alone, increased microglial hippocampal activation and affected cognitive performance in the Morris water maze (MWM), without affecting the cortical areas.⁴² The MWM is a spatial learning task that is dependent on the hippocampus and is unlike the odor- or texture-based digging tasks, like AST, which has been found to be dependent on the integrity of the PFC of rodents.⁸ The deficiencies in hippocampal-dependent tasks and the selective increase in hippocampal microglial activation previously observed in MS180 + TBI adolescents, combined with the surge in hippocampal IL-1β expression observed here, indicate a specific vulnerability of this structure to the effects of the combination of stress and TBI.

The results suggest that the combination of ELS and pediatric mTBI can have serious implications for the long-term outcome of AHT victims. Child abuse and long-term consequences of ELS show an age-dependent effect,^20,81,82 and increase vulnerability to secondary stressors later in life.^19,23–25 Taken as a whole, the evidence suggests that the long-term consequences of AHT could deteriorate further. Supporting this idea are data showing that recovery from pediatric TBI can be influenced by the post-injury family environment.^83,84 This is relevant because social disadvantage – which is common in victims of abuse – in children with TBI predicted more adverse behavioral sequelae and less favorable changes in some outcome measures.⁸³ Moreover, it has been reported that families of children with severe TBI, concurrent limited social resources, and high prevalence of stressors exhibit a higher deterioration of family functioning over time.⁸⁵

Conclusion

Two major conclusions emerge from this study. First, pediatric mTBI causes an impairment in executive function and behavioral flexibility that is related to IL-1β expression in the hippocampus–PFC circuit. Second, although the cognitive impairments are not further aggravated by ELS on individual test stages, ELS rendered an overall elevation in set-loss errors, while the combination of stress and TBI also caused an increase in hippocampal neuroinflammation. The present study provides insight into the complexity of the relationship between stress and TBI observed in victims of AHT, and provides a novel approach that utilizes a pre-clinical model that could more accurately represent the complexity of AHT and facilitate translation to the clinic.

Funding Information

Funding for this study was provided by grants CONACyT-FOSISS No. FIS/IMSS/PROT/1769 (N.L.A.), National Institutes of Health (NIH) grants HD069620, NS099674, and NS084967 (A.E.K.), and NS110609, NS094950, NS099683 (C.O.B.), the University of Pittsburgh Physicians /UPMC Academic Foundation, and the University of Pittsburgh Medical Center (UPMC) Rehabilitation Institute (C.O.B.). A.R. is a graduate student in la Programa de Doctorado en Ciencias Biomédicas- Universidad Nacional Autónoma de México (UNAM) and is supported by CONACYT (CVU no. 509694/286252), Programa Para La Formación de Investigadores de la Coordinación de Investigación en Salud-IMSS, and Programa de apoyo de estudios de posgrado (PAEP-UNAM) to conduct research in the Department of Physical Medicine & Rehabilitation and the Safar Center for Resuscitation Research at the University of Pittsburgh.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Table S1

Supplementary Table S2