All animal procedures conformed to the guidelines issued by the Committee on Animal Research of Peking Union Medical College Hospital (Beijing, China) and were approved by the Institutional Ethics Committee. A total of 38 adult (age, ~10 weeks) male Sprague Dawley^® rats (weight, 227.5±9.6 g) were purchased from Beijing HFK Bioscience Co., Ltd. (Beijing, China). Rats were housed in cages at 15–25°C, in a room with a 12/12-h light/dark cycle. Food and water were available ad libitum. TBI was induced using Feeney's weight-dropping impact method, as previously described (16). The footplate on which the dropped weight impacted was directly attached to the dura and was 4.6×4.0 mm in area. Before the impact, each rat was anesthetized with an intraperitoneal injection of 10% chloral hydrate (400 mg/kg; Sinopharm Chemical Reagent Co., Ltd, Shanghai, China), then a 5-mm diameter bone window was created, with the center located 2.5 mm lateral to the central suture and 1.5 mm behind the bregma. The force of impact to the right hemispherical cortex was 20 g × 25 cm. According to Gao et al (17), this method should cause moderate TBI. All craniotomies and impacts were performed by an experienced researcher within a two-week period.

Evaluation of the modified neurological severity score (mNSS) was performed at post-TBI days 1, 7, 14 and 28. The mNSS is a composite of motor, sensory, balance and reflex tests. Neurological function is graded on a scale of 0–18 (normal score=0; maximal deficit score=18). One point is gained for each specific abnormal behavior or lack of a tested reflex; thus, a higher score indicates more severe injury. This test is suitable for evaluating long-term neurological outcomes after unilateral brain injury (18). Rats were pre-trained for the tests used to determine mNSS for three days before induction of TBI. Evaluation was accomplished by an experienced researcher who was blinded to the grouping. TBI rats were hierarchically divided based on mNSS scores and rats of with the same score were subsequently randomly distributed into an rTMS-treated group (TMS group) and a control group.

High-frequency (5 Hz) rTMS was performed from post-TBI day 2 until the day before rats were sacrificed. A magnetic stimulator (BEMS-1, manufactured by the Institute of Electronic Engineering, Chinese Academy of Science, Beijing, China), a figure-of-eight coil and monophasic pulses were used (19). The coil had seven turns in each loop, with an inner radius of 3.5 cm and an outer radius of 5 cm, which resulted in the whole rat brain being influenced by the induced magnetic field. Stimulation consisted of 36 trains of 25 pulses (each train lasted for 5 sec) with an inter-train interval of 15 sec to prevent the coil from over-heating. A total of 900 pulses were given per day (20). The intensity was set at 60% of maximum output. The coil was placed over the head at a height of ~0.5 cm and was not allowed to contact the scalp to rule out the influence of coil vibration. Careful attention was paid to ensure that maximal stimulation was over the center of the brain. Rats were awake during stimulation and secured in a small blanket. When rTMS was given to the TMS group, rats in the control group were placed in close proximity to the apparatus, so that they could hear the rhythmic sound produced by the coil.

At post-TBI days 7, 14 and 28, three or four rats from each group were deeply anesthetized by an intraperitoneal injection of chloral hydrate (600 mg/kg) and sacrificed by intracardial perfusion with chilled saline, until effluent was colorless, followed by 4% paraformaldehyde in phosphate-buffered saline (PBS; 20 g paraformaldehyde in 500 ml PBS; pH 7.4). Brains were collected and fixed in 4% paraformaldehyde, then stored at 4°C overnight, before being transferred for paraffin embedding. A vibratome (Leica Microsystems GmbH, Wetzlar, Germany) was used to take a series of consecutive 3-µm coronal sections from 1.5 mm behind the bregma (passing through the center of impact). Four sections from each rat were stained with HE, then photographed with a microscope using a ×10 objective (Zeiss AG, Oberkochen, Germany). These sections were taken from a similar position in every rat, determined by the morphology of the lateral ventricle and striatum of the healthy hemisphere. The area of brain parenchyma in each section was evaluated using ImageJ v. 2.1.4.7 analysis software (National Institutes of Health, Bethesda, MA, USA). Relative brain parenchyma loss was then calculated as the percentage difference in parenchyma area between the impacted and non-impacted hemispheres.

In order to detect endogenous cell proliferation, intraperitoneal injection of bromodeoxyuridine (BrdU; 50 mg/kg; Sigma-Aldrich; Merck KGaA, Darmstadt, Germany) was performed immediately after TBI, then every two days until sacrifice (21). Serial coronal sections were taken within 2 mm of the impacted point of the brain. Eight sections (3-µm thick) taken from similar positions in each rat (for example, from across the anterior part of the lateral ventricle) were used for IHC staining. Before immunostaining for BrdU, sections were first incubated in 50% formamide, 2X standard saline citrate for 2 h at 65°C, followed by 2N hydrochloric acid for 30 min at 37°C. The sections were then blocked with 5% donkey serum (Beijing Dingguo Changsheng Biotechnology Co., Ltd., Beijing, China) and 0.3% Triton X-100 in PBS for 30 min at 37°C. Sections were incubated at 4°C overnight with antibodies against BrdU [sheep polyclonal immunoglobulin (Ig)G; 1:200; ab1893; Abcam, Cambridge, UK], musashi1 [a marker for neural stem/progenitor cells (NSPCs); rabbit monoclonal IgG; 1:100; ab52865; Abcam], neuronal nuclei (NeuN; a marker for mature neurons; mouse monoclonal IgG; 1:200; MAB377; EMD Millipore, Billerica, MA, USA) and caspase-3 (a marker for apoptosis; rabbit monoclonal IgG; 1:100; ab179517; Abcam). The specificities of these antibodies were accepted based on the manufacturers' technical information. Following this, sections were rinsed three times with PBS, for 5 min each, and incubated in the dark for 2 h at room temperature with fluorescein isothiocyanate-conjugated donkey anti-chicken (703-095-155), Cy5-conjugated donkey anti-rabbit (711-175-150) or Cy3-conjugated donkey anti-mouse secondary antibodies (711-165-151; all 1:500; all Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA). Subsequently, the sections were rinsed three times, for 5 min each, with PBS. Following this, the sections were incubated at room temperature for 1 h with streptavidin-peroxidase using a VECTASTAIN ABC horseradish peroxidase kit (PK-4004; Vector Laboratories, Inc., Burlingame, CA, USA), rinsed three times with PBS, for 5 min each, and visualized with 0.1 µg/ml 4′,6-diamidino-2-phenylindole (D9542; Sigma-Aldrich; Merck KGaA). The visualizing process was monitored under a microscope. Once sufficiently stained, after 5–10 min, the sections were rinsed with PBS for 10 min. To act as negative staining controls, sections from each rat received identical preparations for IHC staining, except that the primary antibodies were omitted. Sections were observed under a light microscope with an image acquisition system (Leica MC3000; Leica Microsystems GmbH, Wetzlar, Germany) and quantified using ImageJ v. 2.1.4.7 software. Staining of BrdU, NeuN and caspase-3 was assessed by the number of positive cells in a ×200 field, while staining of musashi1 was evaluated by the percentage of musashi1-positive area in total brain parenchyma area in a ×200 field. For BrdU and musashi1, fields with the highest positive cell count were selected from within the dorsolateral subventricular zone (SVZ) (22). Similar fields in the perilesional zone were selected for NeuN and caspase-3.

In order to assess glucose metabolic activity in the perilesional zone, a small number of rats (n=3 per group) were subjected to fludeoxyglucose F 18 (¹⁸F-FDG) micro-PET scanning at post-TBI days 2 and 13. ¹⁸F-FDG was produced with an RDS111 cyclotron and corresponding radiochemical synthesis system (both CTI Molecular Imaging; Siemens AG, Munich, Germany). Rats were anesthetized by intraperitoneal injection of chloral hydrate (400 mg/kg) and injected with 0.5 mCi ¹⁸F-FDG dissolved in 0.4 ml saline into the tail vein. Each animal was scanned with an Inveon dedicated micro-PET system (Siemens AG) following an uptake period of 20 min. The system parameters were set as described by Constantinescu and Mukherjee (23). A 10-min static acquisition was performed in three-dimensional mode, and images were reconstructed using a maximum-a-posteriori probability algorithm. Images acquired from micro-PET were analyzed using ASIPro VM v. 6.0.5.0 software (Concorde Microsystems, Knoxville, TN, USA). A region of interest (ROI) method was used to evaluate change in the regional uptake of PET tracer. The relative metabolic activity was expressed as percentage deficit as compared with the contralateral region for ROIs (the cortex and striatum) in the perilesional zone, as described by Visnyei et al (24).

Data were analyzed using SPSS 19.0 software (IBM SPSS, Armonk, NY, USA). Data are presented as the mean ± standard deviation. An independent-samples t-test was used to evaluate the heterogeneity of baseline data. The difference in unexpected mortality rate between the two groups was evaluated using Fisher's test. For each behavioral test, repeated-measures analysis of variance (ANOVA) was used to test for the inter-group variance as a result of the repeated assessments. The analysis consisted of testing for the group × time interaction. As no group × time interaction was detected, univariant ANOVA was used to access the differences between two groups at all time points. Independent-samples t-tests were also used to evaluate IHC results, relative brain parenchyma loss and relative metabolic activity. P<0.05 was considered to indicate a statistically significant difference.

After weight-dropping induced TBI, 33 rats survived until post-TBI day 2, when TMS commenced. A number of rats (n=8) succumbed unexpectedly before the planned timings; however, the majority (n=25) were sacrificed as planned. The mNSSs at post-TBI day 1 in unexpected mortality and planned sacrifice rats were similar (10.00±2.03 vs. 9.66±1.86, P=0.568). The unexpected mortality rate in the TMS group (8.33%, 1/12) was lower than that of the control group (33.33%, 7/21), but this was not found to be a statistically significant result (Fig. 1).

Rats with an mNSS of 9–14 and surviving for at least 14 days were evaluated for behavioral recovery. The number of rats with a score of 9–14 and surviving for 28 days was too small (n≤2) for statistical analysis, so these data are not shown. On post-TBI day 1, the mNSSs of the TMS group (n=6) and control group (n=5) were similar (Fig. 2; 10.17±1.17 vs. 10.60±1.82, F=0.230, P=0.643), as were the mNSSs of the two groups on day 7 (7.00±1.27 vs. 6.20±1.92, F=0.689, P=0.428). The mNSS of the TMS group on post-TBI day 14 was lower than that of the control group, although this result was not found to be statistically significant (3.38±1.47 vs. 5.40±1.14, F=3.758, P=0.09).

Between post-TBI days 1–28, relative brain parenchyma loss in the impacted hemisphere increased in both groups. As shown in Fig. 3, differences between the TBI group and the control group at post-TBI day 7 and 14 were not significant. At post-TBI day 28, relative brain parenchyma loss in the impacted hemisphere was observed to be lower in the TMS group than in the control group, although this difference was not found to be statistically significant. The typical view of the SVZ corner could not be observed as the brain tissue surrounding the dorsolateral ventricles of the injured hemispheres was necrotic.

In the dorsolateral SVZ, BrdU positive cell density and musashi1 positive area (indicating cell proliferation and NSPCs, respectively) were highest on post-TBI day 14 in both groups (Fig. 4). The BrdU-positive cell density and musashi1 positive area were higher in the TMS group than the control group at day 7, 14 and 28. The BrdU-positive cell density was significantly higher in the TMS group compared with the control group at post-TBI day 14 (175±22 vs. 120±18, P=0.02).

The NeuN-positive cell densities (indicating mature neurons) in the perilesional zones were higher in the TMS group than in the control group at day 7, 14 and 28 (Fig. 5). At day 28, the NeuN-positive cell density was significantly higher in the TMS group compared with the control (143±6 vs. 106±8, P=0.04). Caspase-3 positive cell density (indicating cell apoptosis) in the perilesional zone decreased from day 7 to day 28 in both groups. The caspase-3-positive cell density was significantly lower in the TMS group compared with the control group at post-TBI day 7 (12±6 vs. 31±5, P=0.03) and day 14 (8±4 vs. 20±9, P=0.01).

Compared with the same sites in the non-impacted hemisphere, the metabolic glucose activity in two ROIs (the perilesional zone in the cortex and the striatum of the impacted hemisphere) was reduced on post-TBI day 2 but elevated on post-TBI day 13 in both the TMS and control groups. No significant difference was observed between the two groups (Fig. 6).