Adult (aged 12–14 weeks) male Sprague-Dawley rats were obtained from Charles River Laboratories (Kingston, NY). Upon arrival, rats, were single-housed and maintained on a 12:12h light:dark cycle (0700 lights on) with access to laboratory chow and water ad libitum. Rats were acclimated to the housing conditions for 2 weeks prior to experimentation. Temperatures of housing and preparation rooms were maintained between 22 and 25°C. All experimental procedures occurred between 0700 and 1100 h. The study was approved by the Department of Veterans Affairs New Jersey Health Care System Institutional Animal Care and Use Committee in accordance with AAALAC standards.

An initial ASR test was performed. Rats were matched for amplitudes to the 102 dB stimulus, and then randomly assigned to SHAM and mTBI conditions. The time line of stress, behavioral tests, and sample collection are all relative to the day of injury as post-injury day (PID) and are depicted in Figure 1. The ASR was tested PIDs 1, 3, 7, 14, 21, and 28. Within the SHAM and mTBI groups rats were randomly assigned to one of the following conditions: exposure to the conditioning context chamber (CON group) without shock on PID 34 (SHAM, N = 7; mTBI, N = 7), exposure to a single shock on PID 34 (S34 group; SHAM, N = 8; mTBI, N = 7), exposure to the conditioning chamber on PID 34 and a single shock on PID 35 (S35 group; SHAM, N = 8; mTBI, N = 6), or exposure to a single shock on both PID 34 and PID 35 (2S group; SHAM, N = 8; mTBI, N = 6). To characterize the stress response and its recovery, the two-shock group (SHAM and mTBI) had a basal blood sample at 0700. At 0900, the these groups received a single shock and were returned to their home cages with blood samples obtained through tail nick at 15, 30, and 60 min after shock. For all groups, a terminal sample was obtained 15 min after chamber exposure either on PID 34 or on PID 35.

Fluid percussion injury was induced as previously described (Pang et al., 2015). All rats underwent surgery (−3.0 mm AP, −3.5 mm ML from Bregma; 4 mm diameter) to expose an intact dural surface. Anesthesia for surgery was a mixture of ketamine (60 mg) and xylazine (7 mg) delivered i.p. at 1.0 ml/kg. A leur-lok connector, supported by a section of 12 ml syringe, was inserted into the skull and fixed with dental cement and metal screws. The location of the craniotomy was counterbalanced between right and left hemispheres. The following day, rats were anesthetized with 5% isoflurane (about 60–90 s). A voice-coil controlled device (Wahab et al., 2015) was attached through tubing to the leur-lok connector. Injury was induced when the rat exhibited bilateral foot-pinch reflexes. A single pressure pulse (20.0 ± 0.77 psi peak pressure for mTBI, 0.0 psi for SHAM) was delivered by the device to the cranial surface. Immediately upon delivery of the pressure pulse, apnea (latency to first observable inspiration) and the appearance of tail/limb hyper-extension were observed, then rats were placed supine in an observation cage and righting reflex (latency to righting to prone position on all four limbs) was observed. The injury parameters and immediate responses to injury are contained in Table 1. The degree LFP injury affected apnea, righting reflex, and hyperextension is similar to that of earlier work (Pang et al., 2015).

The apparatus was described previously (Servatius et al., 2008). Foot-shock exposure occurred in an operant chambers (Coulbourn Instruments, Langhorn, PA), enclosed in a sound attenuated box (Coulbourn Instruments, Langhorn, PA), equipped with constant current shockers (Coulbourn Instruments, Langhorn, PA). Graphic State Notation software (v. 3.02, Coulbourn Instruments, Langhorn, PA) controlled the foot-shock delivery. A single 2.0 mA foot-shock (0.5 s duration) was delivered through the grid floor (Coulbourn Instruments, Langhorn, PA) after 1 min in the chamber. Context controls (CON) rats were transferred to the chamber and remained in the chamber for 1 min before transfer back to the home cage.

The acoustic startle reflex (ASR) procedure was conducted as previously described (Servatius et al., 1998; Pang et al., 2015). Rats were allowed to acclimate to the testing apparatus for 10 min prior to the onset of testing. Each test session consisted of 24 presentations of white noise of either 82, 92, or 102 dB (100 ms with 5 ms rise/fall) delivered against a continuous background noise level of 68 dB. Stimuli were delivered in a single pseudorandom order which had the constraint the one each of the three stimulus intensity would be delivered in three consecutive trials with interstimulus intervals of 25–35 s. As in previous work (Pang et al., 2015), a startle response was defined as platform movement within 250 ms after stimulus presentation that significantly exceeded 250 ms baseline movements. The protocol (Servatius et al., 1998; Pang et al., 2015) yields measures of sensitivity (% responses) and responsivity (magnitude of responses; Blumenthal and Goode, 1991) for the three stimulus intensities.

To assess stress reactivity, blood samples were obtained via tail nick as previously described (Servatius et al., 1994). Briefly, blood samples were obtained from a lateral tail vein into heparinized (EDTA contained) capillary tubes (100 ml) in duplicate while the rats were loosely restrained on a sampling table. Individual sampling required <30 s from the time of removal from the home cage to the completion of the duplicate sample. Plasma was separated then stored in non-heparinized capillary tubes at −80°C until assayed for CORT through radioimmunoassay (RIA; CORT I125 RIA kit, MP Biomedicals, Inc., Orangeburg, NY) according to the vendor's instructions. Sacrifice was by rapid decapitation without sedatives 15 min after receiving footshock or placement in experimental context. Trunk blood was collected into serum separator tubes and centrifuged at 2300 rpm for 15 min at 4°C and stored at −80°C until assayed. The brain was extracted and bilateral pre-frontal cortex, bilateral hippocampus and the cerebellum were fresh dissected and frozen at −80°C until assayed. Brain dissection required <30 s. The amount of brain tissue used for assay was 40–70 mg of prefrontal cortex, 60–70 mg for hippocampus, and 250–300 mg for cerebellum.

Quantification was performed by highly sensitive and specific gas chromatography (GC)/mass spectrometry (MS), preceded by high performance liquid chromatography (HPLC) purification as previously described (Marx et al., 2006b,c; Sripada et al., 2013b, 2014), with some modifications. Serum (1.0 mL) was homogenized with 2.0 mL distilled water containing trace tritiated neurosteroid (New England Nuclear) to detect the HPLC fraction of interest, as well as a constant amount of deuterated allopregnanolone (D4-allopregnanolone) and deuterated pregnenolone (D4-pregnenolone) as the internal standards. Supernatants were extracted three times with three volumes of ethyl acetate. Brain tissues were initially suspended in 1.0 mL distilled water and extractions were performed with two volumes of ethyl acetate. HPLC purification was performed on an 1100 Series Agilent instrument. Standards and samples were derivatized utilizing heptafluorobutyric acid anhydride (HFBA) and injected onto an Agilent 5973 mass spectrometer (MS) coupled to an Agilent 6890 N GC equipped with an Agilent HP-5MS 30 m × 0.250 mm × 0.25 μm capillary column. Positive ion electron impact (EI) ionization was utilized in the GC/MS component using helium as the carrier gas. A subset of serum and brain samples were analyzed in duplicate in positive ion EI mode. In addition to the GC/MS retention time characteristic of each neurosteroid, the definitive structural identification of each neurosteroid was provided by its unique mass fragmentation pattern. Mass spectrometer single ion monitoring was used to focus on the most abundant ion fragment for each HFBA derivative [PREG (298.2), ALLO (496.2), and ANDRO (486.2)]. Internal standards consisted of deuterated D4-allopregnanolone for ALLO and ANDRO quantifications, and D4-pregnenolone for PREG quantification (Cambridge Isotope Laboratories, Andover, MA). For neurosteroid quantification, the standard curve for the steroid of interest was prepared by combining known quantities of the neurosteroid (Steraloids, Newport, RI) with a constant amount of deuterated internal standard. Identical to the samples, the standard curve was extracted three times in ethyl acetate prior to HPLC purification and GC/MS injection (standard curve r² = 0.99 for each neurosteroid). The area under the curve of each known quantity of neurosteroid was divided by the area under the peak of the internal standard. This ratio was then plotted on the y-axis against known quantities of each steroid to generate the standard curve. Only peaks with a signal to noise ratio ≥5:1 were integrated. The limit of neurosteroid quantification with this methodology is 1 pg for PREG, ALLO, and ANDRO (femtomolar sensitivity). Intra-assay coefficients of variation were 2.2% for PREG, 3.4% for ALLO, and 4.7% for ANDRO. These 84 intra-assay coefficients of variation were in the same range as our prior investigations utilizing an EI approach.

ASRs were computed as previously reported (Pang et al., 2015). The primary dependent measures are the proportion of responses by stimulus intensity (sensitivity) and the magnitude of responses by stimulus intensity (responsivity). Magnitude was displacement relative to the rat's body weight. These data were analyzed with ANOVA models with repeated measures for Stimulus Intensity and Sessions. Basal CORT levels were assessed separately from CORT stress responses, which were analyzed with a mixed ANOVA. For neurosteroids, multiple analyses of variance (MANOVAs) were computed. For serum, PREG, ALLO, and ANDRO were compared. In the brain, PREG, ALLO, and ANDRO were separately assessed by brain region (prefrontal cortex, hippocampus, and cerebellum). As the equivalent of an F statistic, Hotelling' Trace was used for the test of overall significance (p < 0.05), with Bonferroni t-tests used to understand post-hoc differences. All data are expressed as mean ± standard error of the mean.

Sensitivity to respond decreased in the aftermath of mTBI and did not recover by PID 28. In fact, sensitivity of mTBI rats decreased from PID 14–28 (See Figure 1A). Thus, the suppressive effect of mTBI on ASR sensitivity is long lasting. These impressions were confirmed with a 2 × 3 × 6 (mTBI × Intensity × Session) mixed-ANOVA. The interaction of mTBI × Intensity × Session, F_{(10, 560)} = 2.2, p = 0.012, was significant and superseded all other main effects and interactions. Similarly, mTBI rats displayed ASR magnitudes that were suppressed and which did not change or recover (See Figure 1B). Only the main effects of Intensity, F_{(2, 112)} = 101.4, and TBI, F_{(1, 56)} = 39.5, were significant, ps < 0.001. The dramatic suppression evident in average responding extended to individuals rats in that only 3 of 26 mTBI rats recovered to 50% of their pre-injury ASR magnitude.

CORT levels, derived from basal samples obtained soon after lights on, did not differ between SHAM (4.6 ± 1.2 μg/dl) and mTBI (8.2 ± 2.3 μg/dl) rats, t₍₈₎ = 1.3, p > 0.05. Repeated samples obtained after a single foot shock did not distinguish between SHAM and mTBI rats. This impression was confirmed with a 2 × 3 (TBI × Samples) mixed-ANOVA. Only the time factor, F_{(2, 24)} = 10.4, p < 0.01), was significant (See Figure 2). Essentially, SHAM and mTBI rats were isomorphic in stress response levels and recovery of CORT. Note that CORT levels from the single shock are consistent with those attributable to exposure to a moderately intense stressor.

For serum, a 2 × 4 (mTBI × Stress Group) MANOVA indicated significant effect of Stress Group, F_{(9, 137)} = 1.99, p < 0.05 (See Figure 3). Significant comparisons were indicated for ALLO (CON vs. 2S) and ANDRO (CON vs. S35), all ps < 0.05.

For brain PREG, a 2 × 4 (mTBI × Stress Group) MANOVA indicated an effect of Stress Group, F_{(9, 131)} = 0.013 (See Figure 4). In the prefrontal cortex, PREG levels were elevated in S34 rats compared to CON rats, p < 0.05. In the hippocampus, PREG levels were elevated in S34 rats compared to C and 2S rats, and in S35 rats compared to 2S rats, ps < 0.05. Further, in the hippocampus and cerebellum, PREG levels were elevated in S34 rats compared to 2S rats, ps < 0.05. In the cerebellum, PREG levels were elevated in S34 rats compared to 2S rats.

For brain ALLO, no differences were detected (See Figure 5).

For brain ANDRO, the 2 × 4 (mTBI × Stress Group) MANOVA indicated significant effects of Stress Group, F_{(9, 144)} = 5.8, and mTBI, F_{(3, 46)} = 3.4, ps < 0.05 (See Figure 6). In the prefrontal cortex, the levels of ANDRO of S34, S35 and 2S rats were all elevated compared to CON rats; ANDRO levels of 2S rats were elevated compared to S34 and S35 rats which did not differ, all ps < 0.05. In the hippocampus, ANDRO levels of 2S rats were elevated compared to CON rats, p < 0.05. In the cerebellum, S35 ANDRO levels were elevated compared to CON rats. The ANDRO levels of mTBI rats were higher than those of CON rats in the prefrontal cortex and hippocampus, ps < 0.05. Inasmuch as ANDRO levels of mTBI-CON rats were nominally lower than SHAM-CON rats in both brain regions, elevations are solely from the experience of stress.

Specific a priori comparisons were separately evaluated: (1) whether TBI induced changes in brain neurosteroids in the absence of stress and (2) whether sensitization was apparent and to a greater degree in mTBI rats. MANOVAs of these two specific comparisons were nonsignificant.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.