Establishment of a Rodent Model of Neural Injury After High Frequency Monopolar Stimulation of the Motor Cortex

Direct electrical motor cortex stimulation with short-train high-frequency stimulation (HFS) for motor evoked potentials (MEPs) has been used intraoperatively during supratentorial surgeries, but the safety threshold is poorly dened. The goal of this study is to establish a rat model for the investigation of neural damage in the cerebral cortex caused by high current HFS to aid in dening safety thresholds. We performed bilateral craniotomy on 12 rats. Cerebral sensory-motor cortex was stimulated with a high-frequency current for 100 times. The rats were sacriced and the brains were sliced for Nissl, DAPI, and IBA-1 staining. Severe neural damage of the cerebral cortex was found in all cases, including markedly shrunken and pyknotic cells. IBA-1 staining revealed reactive microglia morphology in the lesion area. DAPI staining showed nucleus degeneration and deformation. The cell density were signicantly lower within the lesion area compared to the contralateral side. This study has established a brain lesion model caused by HFS on rats. These results suggest HFS may carry a risk of serious neural damage if repeatedly applied to the same brain site. More experiments are needed to fully understand the safety threshold of direct cortical stimulation with HFS for clinical use.


Introduction
Intraoperative neurophysiological monitoring (IONM) has been used to reduce the risk of neurological deterioration during neurosurgical procedures [1][2][3][4] . In recent years, short-train high-frequency stimulation (HFS) has been widely used for functional mapping of the motor area and functional monitoring to detect motor dysfunction during intracranial surgery [5][6][7][8][9][10][11][12] . Direct application of electrical stimulation on the motor cortex may carry a risk of damage to the brain. Most previous studies that investigated safety parameters were based on continuously delivered current or prolonged, continuous 50-Hz biphasic rectangular pulse [13][14][15][16][17] . The safety thresholds for direct cortical stimulation with short-train HFS under varying parameters is largely unknown. At present, the current intensity is generally considered to be safe when under 30 mA 9,18 . However, pediatric patients often require higher current to trigger MEPs 19,20 . Under these circumstances, it is necessary to de ne the safe parameters of HFS to ensure that monitoring itself does not injure the patients.
The current safety thresholds used in the clinic for direct cortical stimulation with HFS for motor mapping and monitoring is based on continuously delivered current with low-frequency stimulation (LFS) since there are few studies based on HFS in human or animal models 13-17, 21,22 . A previous study reported that 50-mA current HFS caused mild and transient neural damage in the brain on transmission electron microscope slides 16 . No signi cant neural damage was observed under the light microscope. Thus, to develop a model a higher current HFS is required in order to induce signi cant neural injury. In this study, we used a protocol that goes beyond the typical needs of general clinical use. The primary goal of this study is to de ne characteristics of HFS-injury in the cerebral cortex in an animal model. This will facilitate further experiments to determine safe parameter combinations for HFS.

Stimulation results
Each of the 12 rats received 100 times of short trains of HFS. All rats survived the duration of the experiment and respiration rates remained normal throughout stimulation. The motor evoked potential and muscle twitch were triggered during each stimulation. A summary of stimulation parameters is presented in Table 1.

Gross ndings
After stimulation, the surface of focal dura mater just below and around the stimulated electrodes showed pink-brown color and swelling (Fig. 1B). The average diameter of the lesion on the brain's surface was 4.69 ± 0.56 mm, which was larger than the width of the electrode (3 mm). The surface of the brain was smooth and did not adhere to the dura. During the slicing, the cross section of the brain showed a light-brown colored cone-shaped lesion which had a relatively de ned border (Fig. 1C).

Histological changes of the stimulation sites
The Nissl stained slices were examined under light-microscopy and revealed an edematous area extending through the cortex towards the subcortical corpus callosum and the lateral ventricle in a coneshaped manner (Fig. 1D). There were 8 rats which presented with discrete hemorrhages within the brain parenchyma under the stimulation site (Fig. 1C). In the lesion area, there was increased extracellular space and shrunken cells in contrast to the normal tissue ( Fig. 2A). All lesions showed severe damage and the neurons were markedly shrunken and dark in all grey layers and white matter. The columnar organization of neurons were disrupted in all layers. DAPI staining showed some nuclei were deformed The average area of the lesion on the coronal section and the average volume of lesion with the largest lesion presentation are presented in Table 2. The average thickness of layer I, layer V, corpus callosum, and the entire cortex of the 12 rats were signi cantly larger on the stimulation side than the control side (p < 0.001, < 0.001, < 0.0001, < 0.0001, respectively) (Fig. 2B). The cell density of the lesion (2217.25 ± 248.59 particle/µm 2 ) was signi cantly lower than the control side (2544.21 ± 280.25 particle/µm 2 ) (p < 0.001) (Fig. 1E). The lateral ventricles and corpus callosum were compressed on the lesion side in all 12 cases. The average total cell area within the lesion was 36.77 ± 9.56 µm 2 , which is signi cantly smaller than that of control side (117.93 ± 32.97 µm 2 ) (p < 0.0001) (Fig. 1E). In layer V, most of the motor cells appeared severely shrunken. The ratio of total cell area to total area was 6.42 ± 2.05 % in lesion and 27.41 ± 7.31 % in the corresponding part of the control side.

Behavior assessment
No rats had visible major or minor seizure activity throughout the course of the study. None of the 12 rats showed abnormal EEG waves before or after stimulation. However, the EEG showed a very deep anesthesia pattern from the general anesthesia of iso urane. All rats recovered well from anesthesia and exhibited no motor de cit or abnormal behavior within the 5 hours after stimulation and before sacri ce.

Discussion
In this experimental study, we established an animal model using rats that could simulate the potential brain damage from HFS. We delivered 100 trains of 100 mA HFS to the rat brain. After stimulation, the brains showed swelling and coloring consistent with hemorrhage or hemolysis on visual observation. Comparing Nissl, IBA-1 and DAPI staining from the electrode sites with comparable regions on the contralateral side, it was found that signi cant neural damage was associated with the electrical stimulation. These results indicate that the cortex may be injured by HFS if a certain safety threshold is exceeded.
The safety threshold for transcranial direct cortical stimulation is relatively well established, but has not been clearly identi ed for HFS 13,14,23 . In general, higher current intensity, total charge and total charge density will augment neuronal damage to brain tissue 24  In order to establish a model, we used 100 mA HFS repeated 100 times to observe the potential neural damage. In our model, we didn't remove the thin dura mater because: (1) it is easy to injure the brain when removing it; and (2) the dura mater of the rat can reduce the gap of arachnoid membrane between rat and human brain 33 . In this study, the brain lesion included brain tissue edema, increased extracellular space, severely damaged neural cells, and hemorrhage, corresponding with lesions caused by electrical stimulation 21,26 . According to the grading of neuronal damage as described by Pudenz et al. and Yuen et al., all the rats had severe neural damage 21,26 . The DAPI staining also showed the severe injury of the cell nucleus 34 . IBA-1 staining revealed rami ed or resting microglia on the control side. In the lesion, the microglia showed dense, spherical morphology which was consistent with reactive or phagocytic microglia. It is possible that severe neural damage induced microglial transformation into brain macrophages to remove dead cells within 5 hours of injury 35,36 . It was unclear whether the lesions can be completely repaired over time. If the lesion is severe and cannot be fully repaired, this process may result in glial scars. 37 It remains to be determined whether the damage we observed is transient or converts over time to a typical glial scar.
The charge density and charge per phase are neural excitotoxic cofactors 15,26 . Currently, the safety limits of the HFS technique applied in a short train over a longer period of time remains unde ned. In this pilot study, the charge of one pulse was 50 µC, and the charge density was 707 µC/cm 2 • pulse, which is very high compared to clinical standards. Severe neural damage was observed after 100 trains were delivered within 15 minutes. This nding demonstrates that repeated HFS at the same site may have a cumulative effect and is likely to cause severe neural damage.
Our study has established a rat model for studying neural damage caused by short-train HFS. The current parameters, while exceeding normal clinical standards, caused severe neural damage to the rat brain that are observable and quanti able. Although it may be di cult to apply the safety limits from the animal histologic changes directly to humans, the neural damage observed may cause permanent neural damage in human brain that would increase the potential to induce seizure 38 . The primary effects appear immediately as a direct result of the tissue or cellular injury, while the secondary effects may evolve over a longer period as a result of molecular signaling cascades that are activated by the initial injury. Longer observation post procedure may help de ne behavioral outcomes and whether the changes we observed are transient or persist and lead to scarring.

LIMITATIONS
The purpose of this study was to create an animal model of brain injury caused by direct cortical stimulation with HFS. Therefore, a relatively high current intensity, 100 mA, was used to stimulate. Future studies with varying levels of current intensity are necessary. The stimulating parameter in this study were based on the settings used for human patients with a larger brain volume and may not translate to the smaller rodent brain. The latency of MEPs recorded on rats was very short, so it was di cult to distinguish the MEP response from artifacts. Based on current study, we do not know the long-term effects of the morphological changes induced by the stimulation. Future studies will extend the survival time to observe the brain changes and animal behavior after a longer period and to determine stimulation parameters where damage is rst observed.

Conclusions
This study has established a brain lesion model caused by direct cortical stimulation using HFS on rats. Gross observations, histological and immunohistochemistry methods, such as the Nissl, DAPI and IBA-1 staining were used to identify the injured cells and the phagocytic changes of microglia. Additional experiments are needed to fully de ne the safety threshold of direct cortical stimulation using short-train HFS, and the model established here can be easily replicated by different investigators attempting to study direct cortical HFS.

Animals
All studies were approved by the institutional animal care and use committee at University of California, San Francisco, and whenever possible the ARRIVE guidelines were followed. P23-P30 Sprague-Dawley male rats weighing 70-125 g were purchased from the Charles River Laboratories (Gilroy, CA, USA) and housed on a 12 h reverse light/dark cycle with ad libitum access to food and water.

Anesthesia and surgery protocol
Rats were anaesthetized in an induction chamber with 3.0% iso urane, and then maintained with 1.0% of iso urane in air and oxygen (FiO2 50%) for the surgery. Their heads were rmly xed with ear bars in a stereotaxic frame (David Kopf Instruments, USA). A "U" shape incision was made to expose the skull.
Two craniotomies were performed, one on each hemisphere (3 mm above bregma and 5 mm below bregma with width of 5mm) using a drill. The dura mater was exposed for electrical stimulation (Fig. 1A).

Electrophysiological Stimulation protocol
A Cascade IOMAX (Cadwell Industries, Inc, USA) was used for stimulating and recording. The platinum electrode with 3 mm diameter of a Cortac® subdural strip electrode (PMT Corporation, USA) was xed on the dura mater and used for stimulating. Digitized EEG was recorded during electrical stimulation and 15 minutes after from electrodes placed on the stimulated and contralateral sides. All 12 rats were randomly stimulated on one side of the sensory-motor cerebral cortex. The contralateral side was used as the control side. EMG responses were recorded from identical needle electrodes placed on the triceps brachii and triceps surae muscle. After the stimulation protocol was completed, the skin was closed with suture. All rats were awakened and survived for 5 hours, then were sacri ced for further analysis.

Parameters of electrical stimulation
The stimulus intensity was 100 mA. Complete parameters are listed in Table 1. Charge per pulse(Q) is de ned as I (current intensity) × D (duration of each pulse) for the rectangular pulses. Charge density (QD) (in microcoulumbs/cm 2 , or µC/cm 2 ) is charge divided by electrode area. Total charge (Qt) and total charge density (QDt) are de ned as Q or QD times the number of pulses. 23 Current density (in A/cm 2 ) is applied current at the electrode divided by electrode area.
Brain tissue processing and immunohistochemistry The rats were anesthetized again with iso urane and perfused with 0.01 mol/L phosphate-buffered saline (PBS) (30 mL) and 4% paraformaldehyde (PFA) solution (150 mL). After decapitation, the skulls were opened and brains were carefully dissected. The brains were stored in PFA overnight, then transferred into 30% sucrose. Then they were rapidly frozen and stored in isopentane at -20 ℃. 40 µm thick coronal sections were cut on a Leica CM 1850 cryostat (Leica Microsystems, GmbH, Nussloch, Germany) and mounted onto glass slides. The slides were stained by the Nissl method with cresyl violet and used for quantitative analyses.
Sections were mounted onto glass slides and sealed with Fluoro-Gel mounting media and cover slipped.
Image processing, measurements and cell counts were performed using the FIJI software. 39 The area and volume of each lesion were calculated by its diameter and deepness. Cell counts were obtained by setting the intensity threshold then running the particle count analysis. The cell density was calculated by cell counts divided by area, and the lesion side was contrasted with the control side. The proportion of cell area was calculated by total cell area divided by total area. The thickness of cell layers of cerebral cortex was measured using FIJI software.

Statistical Analysis
Statistical processing and analysis of results were conducted using SPSS (IBM corp., Chicago, USA). The signi cance of differences was assessed using the two tailed t test for independent variables. P values less than 0.05 were considered statistically signi cant. All data were expressed as the mean ± standard deviation.