Repetitive transcranial magnetic stimulation improves neurological function after cerebral ischemia in rats by increasing CREB-regulated TrkB via activation of cAMP/PKA and Ca2+/CaMKIV signaling pathways

Background: Cerebral ischemia is the most prevalent form of clinical stroke. Repetitive transcranial magnetic stimulation (rTMS) can modulate excitability of the cerebral cortex, and this effect is maintained after the stimulation is terminated. However, the underlying mechanisms of rTMS in cerebral ischemia remain unclear. Methods: Herein, we identied the effect of rTMS on cerebral ischemia and further explored the underlying mechanisms. An in vitro model was established using primary cultured neurons under conditions of oxygen-glucose deprivation (OGD), followed by 1 Hz or 10 Hz rTMS treatment. The levels of CREB, PKA and CaMKIV were depleted in neurons to explore the underlying regulatory mechanisms of TrkB by rTMS via CREB. A rat model of cerebral ischemia was established by middle cerebral artery occlusion (MCAO) and the rats were treated with 1 Hz or 10 Hz rTMS to investigate the effect of rTMS on neurobehavior, CREB expression, and cAMP/PKA and Ca 2+ /CaMKIV pathways. Results: rTMS was observed to promote nerve recovery ability in rats with cerebral ischemia, which was accompanied by high expression of TrkB. In OGD-treated neurons, rTMS activated CREB by upregulating cAMP/PKA and Ca 2+ /CaMKIV pathways. Moreover, rTMS induced the activation of CREB to upregulate TrkB. In MCAO rats, rTMS increased the CREB expression, enhanced cAMP, PKA and CaMKIV phosphorylation, and promoted the binding of CREB to TrkB. Conclusions: Taken together, rTMS upregulated CREB and TrkB to improve neurological function in rats with cerebral ischemia by activating cAMP/PKA and Ca 2+ /CaMKIV pathways, which could be of great signicance for cerebral ischemia therapy. area in grayish white color. The ImagemasterVDS image analyzer was used to analyze the percentage of infarct area in the brain tissues to the total brain tissue. Cerebral infarct area ratio (%) = sum of infarct area of all sections/total area of brain × 100%. Triphenyltetrazolium chloride; SAB: Streptavidin-biotin-peroxidase. staining of striatum of 0.05 vs. sham-operated immunouorescence staining of NeuN of MCAO rats under rTMS stimulation 400). Representative of cerebral infarction in the percentage of cerebral infarction area in the cortex of MCAO rats under rTMS stimulation measured by TTC staining. < 0.05 vs. the MCAO rats. The positive expression of TrkB in the cerebral cortical tissues of MCAO rats under rTMS stimulation detected by immunouorescence staining (× 400). * p < 0.05 vs. the MCAO rats. The mRNA expression of TrkB in the cortex of MCAO rats under rTMS stimulation determined by RT-qPCR. * p < 0.05 vs. the MCAO rats. The TrkB protein expression and its phosphorylation in the cortex of MCAO rats under rTMS stimulation determined by Western blot analysis. * p < 0.05 vs. the MCAO rats. The above data were measurement data and expressed as mean ± standard deviation. One-way analysis of variance was used for comparisons among multiple groups, followed by Tukey's post hoc test. n = 5. rTMS, repetitive transcranial magnetic stimulation; TrkB, tropomyosin receptor B; NeuN, neuronal nuclei; Triphenyltetrazolium chloride; MCAO, artery RT-qPCR, reverse transcription quantitative polymerase

promoted cAMP-response element binding protein (CREB) phosphorylation [8]. CREB, which is downstream of various signaling cascades and in the center of multiple kinase pathways driven by activity, could regulate the intrinsic excitability of the neuron and is closely associated with learning and memory abilities [9]. Additionally, CREB triggers the expression of some neuroprotective proteins such as brain-derived neurotrophic factor (BDNF), which facilitates the survival of neurons after ischemia [10]. The activation of CREB regulates the induction of various genes involving BDNF, which serves as a promoter primarily by binding to tropomyosin-related kinase B (TrkB) [11]. As a component of Trk family, TrkB functioned as a receptor tyrosine kinase for BDNF, and both TrkB and BDNF are essential to the nervous system development [12]. Low frequency magnetic stimulation modulates the synaptic plasticity of hippocampal neurons by activating the BDNF-TrkB pathway [13]. A previous study has shown that the activation of the BDNF-TrkB-PI3K/Akt pathway was associated with reduced cell apoptosis in the focal cerebral ischemia [14]. Interestingly, the activation of cAMP/PKA pathway was linked to the brain vascular endothelial cell proliferation after ischemic insult in vitro [15]. Under neuronal stimulation, CaMKIV activation is capable of inducing synaptic modi cations and transcriptional responses, where both of them could affect cognitive and emotional behavior [16]. Based on the above research, rTMS may affect the neurological function after cerebral ischemia in rats through regulating CREB and TrkB. The current study was conducted to investigate the role of rTMS in activating CREB, and the regulation of TrkB in the neurological function of rats with cerebral ischemia. Meanwhile, different pathways such as cAMP/PKA and Ca 2+ /CaMKIV were further explored to offer a better understanding for the underlying molecular mechanisms in development of cerebral ischemia.

Ethics statement
The animal experiments were performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Ethics Committee of North China University of Science and Technology A liated Hospital.

Study subjects
A total of 30 adult male Sprague Dawley (SD) rats (SCXK [Xiang] 2009-0004, aged 8 -10 weeks, weighing 200 -220 g) at speci c pathogen free (SPF) grade were provided by Hunan SLAC Laboratory Animal Co., Ltd. (Changsha, Hunan, China). The rats were allowed free access to food and water with alternate periods of 12 h with and without light. Five rats were regarded as the mock, another 5 were subjected to sham operation and the remaining 20 rats were used for the model establishment.

Construction of adenoviral vector
The adenoviral vector was constructed in 293T cells, using the adenovirus expression vector kit (Takara, Tokyo, Japan). The recombinant adenovirus was obtained at passage 1. Following removal of wild type (WT) virus from 293T cells by differential infection method, adenovirus with a higher titer was collected at passage 4 for later use. siRNA negative control (siNC), siRNA targeting CREB (siCREB), siRNA targeting PKA (siPKA), and siRNA targeting CaMKIV (siCaMKIV) were purchased from Guangzhou Ruibo Biotechnology Co., Ltd. (Guangzhou, Guangdong, China) and packed with the use of adenovirus.

Development of a cell model with OGD and cell transfection
The cerebral cortex, striatum and hippocampus were separated from 24-hour-old neonatal rats, washed 3 times in D-Hanks solution, and cut into pieces. The pieces were detached with 2.5 g/L trypsin at 37°C for 10 min. After the detachment was terminated by Dulbecco's modi ed eagle medium (DMEM) containing 10% fetal bovine serum (FBS), the pieces were centrifuged at 178 × g for 10 min. The stop buffer was removed and the steps were repeated once. The ltrate was then passed through the sieve and centrifuged at 178 × g for 10 min, followed by two washes with D-Hanks solution. Single cell suspensions were prepared by adding DMEM containing 10% bovine serum albumin (BSA) and 10% B-27, and cell density was adjusted to 2 × 10 5 cells/mL. Cells were then plated in 96-well culture plates pre-coated with 0.05 g/L Poly-L-Lysin and cultured at 37°C in a 5% CO 2 incubator. The neurons were cultured in an oxygen-glucose deprivation (OGD) environment (sugar-free DMEM, 5% CO 2 , CO 2 < 1% and 95% N 2 ) for l h every 48 h to prepare single OGD cortical neurons. In order to simulate the model of recurrent ischemia in vivo, two OGD cortical neurons were required to be produced. After receiving the rst OGD treatment for 60 min, neurons were cultured with the conventional culture medium under normal culture conditions for 3 h, and then subjected to OGD treatment for 60 min. After that, the neurons were conventionally cultured. Prior to transfection with adenovirus expressing siNC, siCREB, siPKA and siCaMKIV, primary neurons were seeded in 6-well plates for 48 h. When cell con uence reached 70 -80%, they were transfected according to the instructions described in lipofectamine 2000 (11668-019, Invitrogen, Carlsbad, CA, USA) and cultured for 24 -48 h for subsequent experiments.

Establishment of middle cerebral artery occlusion (MCAO)-induced cerebral ischemia model in rats
Focal cerebral ischemia model was induced by MCAO as previously described [17,18]. General anesthesia was induced with 3% pentobarbital sodium (P3761, Sigma-Aldrich Chemical Company, St Louis, MO, USA) and anesthesia was maintained by inhalation of 1.5% halothane. The rats were partially sterilized, and their skin was cut at the midpoint of the line connecting the left eye and left ear under the microscope. The skin, fascia and salivary glands of the rats were separated and a bipolar coagulator was used to coagulate the blood vessels. The fascia was separated to the tibia, the left tibia was cut, and the muscle tissue was separated to the base of the skull. The skull was removed with a micro-bone drill. The residual skull and the dura mater were removed with a forceps clip. The left middle cerebral artery (L-MCA) was exposed, and the L-MCA was cleaved by a bipolar coagulator at the proximal olfactory bundle. The muscle tissues and salivary glands were reset, and antibiotics were added to the surgical area, wiped with 0.9% sodium chloride solution and sutured. The experimental animals were placed in an incubator after the surgery, and the relevant experiments were performed after the animals were awake. The body temperature of the rats was controlled at 36.5 -37°C during the surgery. In total, 20 rats were successfully operated upon. The sham-operated rats were treated as above but without the electrocoagulation. All the rats survived after surgery. Five of them died within 2 weeks, and 15 rats were successfully modeled. The success rate of modeling reached 75% at 2 weeks. During grouping, there were 5 normal rats, 5 shamoperated rats, 5 MCAO rats, 5 MCAO rats stimulated with 1 Hz rTMS and 5 MCAO rats stimulated with 10 Hz rTMS. rTMS was started 3 days after MCAO-induced ischemic injury [19] as shown in Supplementary   Fig. 1.

Neurobehavioral evaluation of rats
Neurological evaluation was performed on days 1, 3, 5, 7 and 14 following MCAO-induced ischemic injury. The rTMS instrument was produced by Wuhan Yiruide Medical Equipment Co., Ltd. (Wuhan, Hubei, China), with a maximum stimulation frequency of 100 Hz and a circular stimulation coil with diameter of 6 cm. The stimulation intensity was 70% of the maximum output intensity. Primary neurons were placed under a transcranial magnetic stimulator and given low (1 Hz for 900 s) and high (10 Hz for 10 s at an interval of 50 s, total number of pluses of 900) frequency stimulation. The rats used for control were also placed under the stimulation device but without stimulation. After the end of the treatment, the cells were further incubated for 3 h for subsequent experiments. After inducing for 24 h, the rats in the rTMS groups were treated with rTMS once a day for 2 weeks at low (1 Hz for 900 s) and high frequency (10 Hz for 10 s at an interval of 50 s, total number of pluses was 900) conditions. All the rats were given free access to food and water during the experiment.The neurological evaluation of rats was performed by two informed observers based on the following standards: (1) score with the reference to the Bederson neurological de cit, with the score range of 0 -3; the higher score indicated the more severe neurological de cit; (2) score of the wire gripping ability with the score range of 0 -3; (3) learning and memory ability test: the rats were tested for learning and memory ability by Y-type electrical maze after rTMS treatment, and the number of correct responses in each group was recorded. Each group of rats was tested 3 times for averaging.On the 14 th d after injury, all rats were injected with 2% pentobarbital sodium and ketorolac tromethamine to relieve pain. The left ventricle was perfused with 200 mL 0.9% sodium chloride solution. The perfused rats were immediately euthanized with CO 2 inhalation, and the brain was extracted and stored at -70°C for future use.
Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) A total of 1 mL Trizol reagent (Invitrogen, Carlsbad, CA, USA) was added to samples while they were simultaneously pulverized in an ice bath. Total RNA was extracted using an RNA extraction kit (Qiagen company, Hilden, Germany) for real-time PCR. The RNA was reverse transcribed into cDNA (50 ng/μL) using the PrimeScript TM RT reagent Kit (RR047A, Beijing Zhijie Fangyuan Technology Co., Ltd., Beijing, China). Primers (Table 1) were designed using primer Premier 5.0 software and synthesized by Beijing Qingke Biotechnology Co., Ltd. (Beijing, China). RT-qPCR was performed on the ABI 7900HT real-time quantitative PCR instrument (ABI 7900, Shanghai Pudi Biotechnology Co., Ltd., Shanghai, China) using a two-step method, with β-actin as an internal reference. The TrkB mRNA expression was determined by the 2 -ΔΔCt method. Three replicate wells were set for each gene of each sample. The experiment was repeated three times.
Chromatin immunoprecipitation (ChIP) The cerebral cortical tissue was cut into 2-mm pieces, cross-linked in 1% formaldehyde and stored at -80°C for later use. The nucleus was extracted with microtip, lysed, and sonicated until the DNA fragment was about 200 -1000 bp in length. Next, 40 μL Protein G beads and 3 μg antibody were pre-incubated together for at least 1 h prior to overnight incubation with the DNA fragment for immunoprecipitation. The DNA/protein complex was eluted from the beads and reversely cross-linked overnight at 65°C. The DNA was detached by ribonuclease A and proteinase K, followed by phenol/chloroform extraction and alcohol precipitation. Real-time quantitative PCR was performed using iQ SYBR Green Supermix (Bio-Rad Laboratories, Shanghai, China). The amount of immunoprecipitated DNA was calculated by comparison with the total amount of input DNA. Data obtained represent four independent ampli cations. ChIP was carried out with CREB antibody, and qPCR quanti ed the exon of the precipitated TrkB at CRE sites.

Fractionation of cytoplasmic and nuclear proteins
Cytoplasmic and nuclear fractions were isolated according to the instructions of the Nuclear Protein Extraction Kit (C500009-0050, Shanghai Sangon Biotech Co., Ltd., Shanghai, China). The internal references of nuclear and cytoplasmic proteins were lamin A (ab26300) and GAPDH (ab9485), respectively.

Co-Immunoprecipitation
The cells or tissues were collected for co-immunoprecipitation. A part of the fragmented chromatin was taken as an input before immunoprecipitation. The cells or tissues were precipitated with polyclonal antibody speci c to TrkB (ab18987, 1 : 1000, Abcam Inc., Cambridge, MA, USA), washed two times with pre-cooled PBS, and fully lysed. After centrifuged at 25764 × g for 3 -5 min, the supernatant was collected and probed with primary antibody at 4°C overnight. Then, 50% agarose beads at a volume ratio of l:10 were added and fully mixed at 4°C for 10 min. After centrifugation for 15 min at 118 × g and 4°C, the supernatant was discarded and the agarose beads remained. After the last wash, SDS protein loading buffer was added to the cells or tissues and then they were mixed, boiled for 5 min, and centrifuged at 25764 × g for 3 min. The supernatant was then collected for SDS-PAGE electrophoresis analysis. Primary antibodies (Abcam Inc., Cambridge, UK) to phospholipase C gamma 1 (PLC-γ1; ab76155, 1 : 5000), shc (ab33770, 1 : 1000) and N-methyl-D-aspartate receptor subunit NR1 (NMDAR-NR1; ab134308, 1 : 1000) were used for Western blot analysis. The non-speci c rat IgG at 1 : 50 was added as NC. The above experiments were repeated 3 times.

Detection of intracellular Ca 2+ concentration
Ca2 + uorescent probe Fura-2/AM was mixed with the calcium-containing solution at a ratio of 1:499, and then added into each group of cells. The cells were then incubated at 37°C and de-esteri ed. The uorescence of Fura-2/AM was excited using excitation light at wavelengths of 340 and 380 nm. The dynamic change of uorescence was monitored by Charge coupled device (CCD) and analyzed by the calcium uorescence imaging system IPA software. The change value of the uorescence intensity ratio of the two excitation lights (Δratio) re ected the intracellular calcium ion concentration [Ca 2+ ]i.

Enzyme-linked immunoassay (ELISA)
A total of 100 pL enzyme immunoassay (EIA) buffer was added to the non-speci c binding wells, and 50 μL EIA Buffer was added to the largest amount of binding well. Next, 50 μL buffer was pipetted from the No. 8 tube into the two standard wells of the lowest sections, and another 50 μL buffer was pipetted from the No. 7 tube into the next two standard wells. This procedure was continued until all standards were added into the standard wells. In addition to the total activity (TA) and blank wells, 50 μL cAMP acetylcholinesterase tracer was added to each well. In addition to TA, nonspeci c binding (NSB), and blank wells, 50 μL cAMP EIA antiserum was added to each well. The plate was covered with plastic lm and incubated at 4°C for 18 h. The liquid in the wells was discarded, and the wells were washed ve times with wash buffer, followed by the addition of 200 μL Ellman reagent into each well. Afterwards, 5 pL of tracer agent was added into the TA well. The plate was then covered with plastic lm, wrapped with tin foil paper, and shaken for 2 h on the shaker. The multi-function microplate reader was used to read the optical density (OD) value at 412 nm.
Triphenyltetrazolium chloride (TTC) staining The fresh brain tissues were placed in a refrigerator at -20°C for 10 min, and sliced into 1-mm sections. The sections were incubated with PBS containing 1% TTC (pH = 7.4) in a 37°C water bath for 20 min. The staining solution was shaken to ensure adequate contact with brain tissue. After the staining was terminated, the sections were xed in 10% formaldehyde overnight. The viable brain tissue was stained in brick red color, and the infarct area in grayish white color. The ImagemasterVDS image analyzer was used to analyze the percentage of infarct area in the brain tissues to the total brain tissue. Cerebral infarct area ratio (%) = sum of infarct area of all sections/total area of brain × 100%.

Immuno uorescence
Cortex, hippocampus and striatum were extracted after the experimental animals were xed, perfused and dehydrated. Afterwards, 30 µM-thick brain sections were made using a freezing microtome. After non-speci c blocking, the sections were incubated in antibody solution containing anti-rat synaptophysin (Millipore, Billerica, MA, USA) or anti-rabbit PSD-95 (Millipore, Billerica, MA, USA) at 4°C for 24 h. After being washed with PBS, the sections were incubated with anti-Alexa-488 and 568-labeled secondary antibodies, respectively. The sections were then placed under a Nikon uorescence microscope (TE2000, Nikon, Tokyo, Japan) to observe the staining intensity. The average uorescence intensity was determined by the ImageJ software package. Immunohistochemical stainingThe rat cortex was xed in 10% neutral formalin solution, dehydrated for 24 h with gradient ethanol, embedded in para n for 12 h, and cut into 5 slices with a thickness of about 3 -4 μm. The sections were dewaxed with xylene I and II for 10 min respectively, dehydrated by gradient ethanol (100%, 95%, 80% and 70% for 2 min respectively), and then washed twice with PBS (5 min for each time). The sections were then soaked in 3% H 2 O 2 for 10 min and washed twice with PBS (5 min for each time). After antigen retrieval under high-pressure for 90 s, the sections were cooled at room temperature and incubated with 5% BSA blocking solution at 37°C for 30 min. Subsequently, the sections were incubated with 50 μL rabbit anti-mouse antibody to CREB (ab32515; 1 : 100) at 4°C overnight, followed by incubation with 50 μL biotinylated mouse anti-goat IgG

Statistical analysis
All data were statistically analyzed using a Statistic Package for Social Science (SPSS) 21.0 (IBM Corp. Armonk, NY, USA). The measurement data were expressed as mean ± standard deviation. The data of multiple groups obeying the normal distribution and homogeneity of variance were compared using oneway analysis of variance (ANOVA) with Tukey's post-hoc test, and data at different time points were compared using repeated measurement analysis of variance. The difference was statistically signi cant at p < 0.05.

Results
rTMS enhanced the nerve recovery ability of MCAO rats and increased TrkB expression The rat model with cerebral ischemia was developed using MCAO. Neurobehavioral evaluation at days 1, 3, 7 and 14 after injury and NeuN immuno uorescence staining were conducted to con rm the successful establishment of the rat model. As demonstrated in Table 2, MCAO rats were weaker than normal and sham-operated rats in terms of neurological function, motion balance, and learning and memory function (p < 0.05). It was found that the uorescence intensities of NeuN in the cortex, hippocampus and striatum of MCAO rats were signi cantly lower than that in the normal and sham-operated rats (p < 0.05, Fig. 1A).
The above results suggested that the rat model of permanent focal cerebral ischemia was successfully established, which was characterized by most obvious cortical changes.
In order to evaluate whether rTMS stimulation could improve the neurological function of MCAO rats, the neurobehavioral changes in rats were examined at days 1, 3, 7 and 14 after injury. The results showed that both 1 Hz and 10 Hz rTMS increased the neurological function, motion balance, learning and memory function of MCAO rats (Table 3). NeuN immuno uorescence of cortex demonstrated that the uorescence intensity of NeuN in the MCAO rats was signi cantly increased following the stimulation of 1 Hz or 10 Hz rTMS (Fig. 1B). The infarct volume in brain tissue was detected by TTC staining, which showed that the infarct volume of brain tissue was decreased in the MCAO rats after stimulation with 1 Hz or 10 Hz rTMS (Fig. 1C). The immunohistochemical detection of TrkB revealed that after MCAO model establishment, either 1 Hz or 10 Hz rTMS increased the positive expression of TrkB in MCAO rats (Fig.  1D). RT-qPCR and Western blot analysis illustrated that the expression of TrkB was upregulated and the TrkB phosphorylation in the MCAO rats was enhanced following stimulation with 1 Hz or 10 Hz rTMS ( Fig. 1E-F). The above results demonstrated that rTMS could enhance the neurological recovery and increase the expression of TrkB in rats with cerebral ischemia. rTMS activates CREB via the cAMP/PKA pathway The primary neurons of the cerebral cortex were treated with OGD and placed under a transcranial magnetic stimulator to receive low (1 Hz) and high frequency (10 Hz) stimulation. The protein expression of CREB and PKA and their phosphorylation levels were measured by Western blot analysis. As depicted in Fig. 2A and B, compared with the OGD-treated neurons, the expression of CREB and PKA were unchanged, and their phosphorylation levels were signi cantly up-regulated in the OGD-treated neurons after stimulation with 1 Hz or 10 Hz rTMS. The results of ELISA showed that the cAMP expression was increased by stimulation with 1 Hz or 10 Hz rTMS in the OGD-treated neurons (Fig. 2C). The results of immuno uorescence in Fig. 2D illustrated that compared with the OGD-treated neurons without stimulation, the uorescence intensity of PKA was signi cantly increased in the nucleus of the OGDtreated neurons stimulated with 1 Hz or 10 Hz, but signi cantly decreased in the cytoplasm (p < 0.05).
After nuclear and cytoplasmic protein separation, the expression of PKA, GAPDH, and Lamin A protein in the nucleus and cytoplasm were analyzed by Western blot analysis. The results revealed that the OGDtreated neurons had increased PKA protein expression in the nucleus but decreased PKA protein expression in the cytoplasm after stimulation of 1 Hz or 10 Hz rTMS (Fig. 2E). According to the coimmunoprecipitation and Western blot analysis, compared with the OGD-treated neurons without stimulation, PKA antibody in the OGD-treated neurons stimulated with 1 Hz or 10 Hz bound to more CREB (p < 0.05, Fig. 2F). When the expression of PKA was down-regulated, the expression of CREB and its phosphorylation were determined by Western blot analysis. The results showed that after 1 Hz or 10 Hz rTMs, the expression of CREB was unchanged, while its phosphorylation was signi cantly decreased in the OGD-treated neurons by the infection of adenovirus expressing siPKA (Fig. 2G). These results suggested that rTMS increased the expression of cAMP, CREB and PKA phosphorylation in primary cultured cortical neurons, stimulated the translocation of PKA into nuclei, as well as enhanced the interaction between PKA and CREB, while siPKA could inhibit the extent of CREB phosphorylation. As described above, rTMS activated CREB via the activation of the cAMP/PKA pathway. rTMS activates CREB via activating the Ca 2+ /CaMKIV pathway rTMS has been demonstrated to increase the extent of CREB phosphorylation. OGD treatment was also performed on primary cortical neurons which were then placed under the transcranial magnetic stimulator, respectively with low (1 Hz) and high frequency (10 Hz) stimulation. The uorescent probe Fura-2 was used to determine intracellular Ca 2+ concentration. The results showed (Fig. 3A) that when compared with OGD-treated neurons without stimulation, [Ca2+]i was signi cantly higher in OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS (p < 0.05). Western blot analysis showed that the expression of CaMKIV was unchanged, and its phosphorylation was signi cantly higher in OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS when compared with OGD-treated neurons without stimulation (p < 0.05; Fig. 3B). After CaMKIV expression was down-regulated, the expression of CREB and its phosphorylation were measured by Western blot analysis. As shown in Fig. 3C, siCaMKIV inhibited CREB phosphorylation in OGD neurons treated by 1 Hz or 10 Hz rTMS. These results indicated that 1 Hz or 10 Hz rTMS potentiated CREB phosphorylation by activating the Ca2+/CaMKIV pathway.

rTMS upregulates TrkB by activating CREB
The primary neurons of cerebral cortex were treated with OGD and placed under a transcranial magnetic stimulator, and given low (1 Hz) and high frequency (10 Hz) stimulation respectively. The results of RT-qPCR of TrkB mRNA expression showed that, when compared with the OGD-treated neurons without stimulation, the mRNA expression of TrkB was signi cantly higher in the OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS (p < 0.05; Fig. 4A). The protein expression of TrkB and its phosphorylation determined by Western blot analysis revealed that OGD neurons had unchanged expression of TrkB but enhanced TrkB phosphorylation with the stimulation of 1 Hz or 10 Hz rTMS (Fig. 4B). The binding of CREB to TrkB at CRE sites in ChIP assay is shown in Fig. 4C. Compared with OGD-treated neurons without stimulation, enrichment of TrkB at CRE sites in OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS was signi cantly higher (p < 0.05). After CREB was down-regulated, the RT-qPCR determination of TrkB mRNA expression showed that the treatment of siCREB decreased the mRNA expression of TrkB in OGDtreated neurons stimulated with 1 Hz or 10 Hz rTMS (Fig. 4D). Western blot analysis was used to determine the expression of TrkB and its phosphorylation (Fig. 4E), and the results showed that siCREB decreased the extent of TrkB phosphorylation in OGD neurons induced by 1 Hz or 10 Hz rTMS with unchanged expression of TrkB. Co-immunoprecipitation and Western blot analysis were employed to characterize the expression of PLC-γ1, shc and NMADR-NR1 co-immunoprecipitated with TrkB (Fig. 4F). The results demonstrated that when compared with OGD-treated neurons, the expression of PLC-γ1, shc and NMADR-NR1 co-immunoprecipitated with TrkB was signi cantly increased in OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS (p < 0.05). After CREB was down-regulated, the expression of PLC-γ1, SHC and NMADR-NR1 co-immunoprecipitated with TrkB was signi cantly decreased in OGD-treated neurons stimulated with 1 Hz or 10 Hz rTMS (Fig. 4G). These results indicate that rTMS could increase the TrkB expression, enhance the binding of CREB to TrkB at CRE sites, and promote the recruitment ability of TrkB, while siCREB could inhibit the TrkB expression. In summary, rTMS increased the TrkB expression by activating CREB.

rTMS promotes the binding of CREB to TrkB in vivo
The effect of rTMS on the interaction between CREB and TrkB was assessed in MCAO rats. The protein expression of CREB and its phosphorylation were determined by Western blot analysis (Fig. 5A). When compared with the MCAO rats, the expression of CREB in MCAO rats stimulated with 1 Hz or 10 Hz rTMS was unchanged, and the extent of CREB phosphorylation was signi cantly increased (p < 0.05). The results of immunohistochemical detection, the positive rate of phosphorylated CREB protein in the MCAO rats stimulated with 1 Hz or 10 Hz rTMS was signi cantly higher than that in the MCAO rats without stimulation (p < 0.05; Fig. 5B). The cAMP expression measured by ELISA showed that 1 Hz or 10Hz rTMS both increased the cAMP expression in the MCAO rats (Fig. 5C). Compared with the MCAO rats, the expression of PKA in the MCAO rats stimulated with 1 Hz or 10 Hz rTMS was unchanged, while the extent of its phosphorylation was notably increased (p < 0.05; Fig. 5D). The expression of CaMKIV and its phosphorylation showed changes consistent with PKA (p < 0.05; Fig. 5E). The co-immunoprecipitation and Western blot analysis demonstrated that the amount of TrkB co-immunoprecipitated with CREB was signi cantly higher in the MCAO rats stimulated with 1 Hz or 10 Hz rTMS than in the MCAO rats without stimulation (p < 0.05; Fig. 5F). The above results suggested that rTMS could increase the expression of cAMP and enhance CREB, PKA and CaMKIV phosphorylation, hence promoting the binding of CREB to TrkB in vivo. Discussion rTMS has been reported to regulate progression in the human brain, and was associated with blood ow and oxygenation changes in the cerebral cortex, suggesting its potential role as a therapeutic technique for neurologic conditions [20]. Therefore, this study investigated the speci c roles of rTMS in neural function after cerebral ischemia in rats. Collectively, the present study has demonstrated that rTMS could improve neurological function after cerebral ischemic in rats by increasing the expression of TrkB via activation of the CREB.
By establishing the OGD neuron model and the MCAO rat model, this study discovered that rTMS could enhance the nerve recovery ability of rats with cerebral ischemia, accompanied by high expression of TrkB. rTMS functions as a nonsurgical technique for cerebral stimulation with promising effects and causes changes in neuroplasticity, thus holding therapeutic potential for neurological diseases [21]. rTMS over motor cortex could promote recovery, which may be bene cial in acute stroke therapy [22]. In addition, rTMS could exert its function on the reduction of unilateral spatial neglect and improve upper extremity function after subacute stroke, subsequently stimulating neuron depolarization and changing the excitability within the cerebral cortex after rMS [23]. rTMS with low intensity elicits the activation of BDNF and TrkB, which further inhibits the dysfunction of spatial cognition in aging mice [24]. A smallmolecule TrkB agonist 7,8-dihydroxy avone has been reported to alleviate cerebral ischemia and reperfusion injury in rats [25]. Hence, the neurological function of rats with cerebral ischemia could be improved following treatment with rTMS and upregulation of TrkB, which was consistent with the results of this study.
In addition, our study demonstrated that rTMS could activate CREB through the cAMP/PKA and Ca 2+ /CaMKIV pathways by elevating [Ca 2+ ]i, and the extent of CREB, cAMP, PKA and CaMKIV phosphorylation. cAMP is an intracellular second messenger that exerts various physiological effects through activation of PKA where the cAMP/PKA pathway is associated with neural control of glucose homeostasis in mammals [26]. The cAMP/PKA pathway has been reported to play a signi cant role in the transcriptional activities of CREB [27]. Based on the data on neuronal differentiation from a study by Ca 2+ /CaMKK, a novel element of the caMK family, could induce the activation and phosphorylation of speci c downstream protein kinases to trigger various Ca 2+ -signaling pathways, which is vital for modulation of gene expression regulated by CREB phosphorylation [28]. The Ca 2+ /CaMKIV pathway has been linked to the activation of transcription factors such as CREB [29]. In addition, Ca 2+ -dependent CREB/c-fos activation by Ca 2+ -CaMKIV triggers osteoclast-speci c gene regulation at transcriptional level through NFATcl [30]. Moreover, the high-frequency rTMS caused CREB phosphorylation and transcription of BDNF through the activation of the Ca 2+ -CaMKII-CREB pathway in the Neuro-2a cells [31]. As a member of the CaMK family, Ca 2+ /CaMKIV along with its protein product was signi cantly detected in the nucleus, which modulates gene transcription by CREB phosphorylation and activation [32]. From the above reported studies, it could be concluded that rTMS induced the activation of CREB via the cAMP/PKA and Ca 2+ /CaMKIV pathways.
Another signi cant nding in this study was that rTMS could increase the TrkB expression through activating CREB. A previous study has demonstrated that magnetic stimulation with low frequency could modulate structural synaptic plasticity of hippocampal neurons by activating BDNF-TrkB pathways [13]. As a member of the neurotrophin family, BDNF improves the functional recovery and triggers neuroprotective effects after ischemic stroke, and its protective effects are exerted by binding to the higha nity TrkB receptor [33]. It has also been demonstrated that daily 5 Hz rTMS for 5 d enhanced the BDNF-TrkB pathway in both cortex and lymphocytes by elevating the demand of BDNF for TrkB [34]. Also, the CREB activation could result in increased expression of anti-apoptotic protein Bcl-2, and is conducive to neuronal survival after ischemic attack [35]. Overall, the aforementioned ndings are supportive of our results where rTMS activated CREB and TrkB, thereby improving neurological function under cerebral ischemic conditions.

Conclusions
In brief, rTMS could contribute to the protection of patients against cerebral ischemia by improving the neurological function via activating CREB, to regulate TrkB through the cAMP/PKA and Ca 2+ /CaMKIV pathways (Fig. 6). This nding suggests that rTMS is a promising therapeutic module for patients who have suffered from cerebral ischemia. Nevertheless, in-depth investigations on the function of rTMS in cerebral ischemia are ongoing and further exploration of the underlying mechanisms of rTMS is

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.  Figure 1 The nerve recovery ability is promoted and TrkB expression is increased by rTMS in cerebral ischemia rats. A), The immuno uorescence staining of NeuN in the cortex, hippocampus and striatum of rats after  after inhibition of PKA determined by Western blot analysis. * p < 0.05 vs. the OGD-treated neurons stimulated with 1 Hz rTMS, # p < 0.05 vs. the OGD-treated neurons stimulated with 10 Hz rTMS. The above data were measurement data and expressed as mean ± standard deviation. One-way analysis of variance was used for comparison among multiple groups, followed by Tukey's post hoc test. n = 5. rTMS, repetitive transcranial magnetic stimulation; CREB, cAMP-response element binding protein; PKA, protein kinase A; OGD, oxygen-glucose deprivation; ELISA, Enzyme-linked immunoassay; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.   with TrkB measured by Western blot analysis, * p < 0.05 vs. the OGD-treated neurons. G), The expression of PLC-γ1, SHC and NMADR-NR1 co-immunoprecipitated with TrkB measured by Western blot analysis after downregulation of CREB, * p < 0.05 vs. the OGD-treated neurons stimulated with 1 Hz rTMS, # p < 0.05 vs. the OGD-treated neurons stimulated with 10 Hz rTMS. The above data were measurement data and expressed as mean ± standard deviation. One-way analysis of variance was used for comparison among multiple groups, followed by Tukey's post hoc test, n = 5. rTMS, repetitive transcranial magnetic stimulation; TrkB, tropomyosin receptor kinase B; CREB, cAMP-response element binding protein; RT-qPCR, reverse transcription quantitative polymerase chain reaction; OGD, oxygenglucose deprivation. the MCAO rats. The above data were measurement data and expressed as mean ± standard deviation.
One-way analysis of variance was used for comparison among multiple groups, followed by Tukey's post hoc test, n = 5. CREB, cAMP-response element binding protein; MCAO, middle cerebral artery occlusion; ELISA, Enzyme-linked immunoassay; PKA, protein kinase A; TrkB, tropomyosin receptor kinase B.